diff --git a/.editorconfig b/.editorconfig
new file mode 100644
index 0000000..7a7bb15
--- /dev/null
+++ b/.editorconfig
@@ -0,0 +1,14 @@
+# Editor configuration, see http://editorconfig.org
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+indent_style = space
+indent_size = 2
+insert_final_newline = true
+trim_trailing_whitespace = true
+
+[*.md]
+max_line_length = off
+trim_trailing_whitespace = false
diff --git a/.gitattributes b/.gitattributes
new file mode 100644
index 0000000..dfe0770
--- /dev/null
+++ b/.gitattributes
@@ -0,0 +1,2 @@
+# Auto detect text files and perform LF normalization
+* text=auto
diff --git a/LICENSE b/LICENSE
new file mode 100644
index 0000000..887bd4e
--- /dev/null
+++ b/LICENSE
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2018 José M. Moreno <josemmo@pm.me> (https://github.com/josemmo)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..fc97c93
--- /dev/null
+++ b/README.md
@@ -0,0 +1,10 @@
+# Hotdog or Not Hotdog
+![Lighthouse PWA: 100/100](https://lighthouse-badge.appspot.com/?score=100&category=PWA)
+![Lighthouse Performance: 93/100](https://lighthouse-badge.appspot.com/?score=93&category=Perf)
+
+Yet another sample project using [TensorFlow.js](https://js.tensorflow.js/) and MobileNet to create a Progressive Web App (PWA) which recognizes Hotdogs and Not Hotdogs, as shown in the HBO series' Silicon Valley.
+
+[![View video](https://img.youtube.com/vi/ACmydtFDTGs/mqdefault.jpg)](https://www.youtube.com/watch?v=ACmydtFDTGs)
+
+## License
+This project in under the MIT License. You can freely do whatever you want with its code as long as you credit the author.
diff --git a/app.css b/app.css
new file mode 100644
index 0000000..ad392d0
--- /dev/null
+++ b/app.css
@@ -0,0 +1,98 @@
+/* COMMON */
+@font-face {
+  font-family: 'Roboto';
+  font-style: normal;
+  font-weight: 900;
+  src: url('roboto-black.ttf') format('truetype');
+}
+body, button {
+  font-family: 'Roboto', sans-serif;
+  font-weight: 900;
+}
+body { background: #202020 }
+* { user-select: none }
+
+
+/* CAMERA */
+#camera {
+  position: absolute;
+  left: 0;
+  top: 0;
+  width: 100%;
+  height: 100%;
+  object-fit: cover;
+  z-index: 1;
+}
+
+
+/* OVERLAY */
+#overlay {
+  position: absolute;
+  left: 0;
+  top: 0;
+  width: 100%;
+  height: 100%;
+  z-index: 2;
+}
+
+#overlay .dialog {
+  position: absolute;
+  left: 30px;
+  right: 30px;
+  top: 50%;
+  transform: translateY(-50%);
+  padding: 30px;
+  background: #21a8c6;
+  color: #fff;
+  border: solid 2px;
+  border-radius: 8px;
+  text-align: center;
+  font-size: 18px;
+}
+
+#overlay .result {
+  position: absolute;
+  left: 0;
+  right: 0;
+  top: 0;
+  padding: 15px 0;
+  color: #ff0;
+  -webkit-text-stroke: 1.7px #000;
+  filter: drop-shadow(-2px 1px 0 #fff) drop-shadow(2px 1px 0 #fff);
+  text-align: center;
+  font-size: 40px;
+  display: none;
+}
+#overlay .result::after {
+  content: '';
+  box-sizing: border-box;
+  position: absolute;
+  top: calc(100% - 20px);
+  left: 50%;
+  margin-left: -50px;
+  width: 100px;
+  height: 100px;
+  background: url('images/hotdog.png') no-repeat center;
+  background-size: 80%;
+  border-radius: 50%;
+  z-index: -1;
+}
+#overlay .result.hotdog,
+#overlay .result.hotdog::after {
+  background-color: #0f0;
+}
+#overlay .result.not-hotdog,
+#overlay .result.not-hotdog::after {
+  background-color: #f00;
+}
+#overlay .result.not-hotdog::before {
+  content: '';
+  position: absolute;
+  top: calc(100% - 20px);
+  left: 50%;
+  margin-left: -50px;
+  width: 100px;
+  height: 100px;
+  background: url('images/cross.png') no-repeat center;
+  background-size: 70%;
+}
diff --git a/app.js b/app.js
new file mode 100644
index 0000000..5215cbb
--- /dev/null
+++ b/app.js
@@ -0,0 +1,214 @@
+(function() {
+
+  var MODEL_PATH = 'model/mobilenet.json';
+  var CLASSES_PATH = 'model/mobilenet.classes.json';
+  var IMAGE_SIZE = 224; // Required by this model
+  var PREDICT_INTERVAL = 1000; // Delay between predictions in milliseconds
+  var MAX_CLASSES = 8;
+
+  var overlayDialog = document.querySelector('#overlay .dialog');
+  var camera = document.getElementById('camera');
+  var notification = new Audio('notification.ogg');
+  notification.preload = 'auto';
+
+  var model, modelClasses;
+  var lastCheck = null;
+
+
+  /**
+   * Update UI
+   * @param {Boolean} isHotdog Is hotdog
+   */
+  function updateUI(isHotdog) {
+    if (lastCheck == isHotdog) return;
+
+    // Update result overlay
+    var result = document.querySelector('#overlay .result');
+    if (isHotdog) {
+      result.classList.remove('not-hotdog');
+      result.classList.add('hotdog');
+      result.innerHTML = 'Hotdog!';
+
+      // Play sound
+      notification.play();
+    } else {
+      result.classList.remove('hotdog');
+      result.classList.add('not-hotdog');
+      result.innerHTML = 'Not Hotdog!';
+    }
+    result.style.display = 'block';
+
+    // Update last state
+    lastCheck = isHotdog;
+  }
+
+
+  /**
+   * Get camera image
+   * @return {HTMLCanvasElement} Canvas with image
+   */
+  function getCameraImage() {
+    var c = document.createElement('canvas');
+    c.width = IMAGE_SIZE;
+    c.height = IMAGE_SIZE;
+    c.getContext('2d').drawImage(camera, 0, 0, c.width, c.height);
+    return c;
+  }
+
+
+  /**
+   * Connect to camera
+   * @async
+   * @return {Promise} Callback
+   */
+  function connectToCamera() {
+    return new Promise(function(resolve, reject) {
+      navigator.mediaDevices.getUserMedia({
+        video: {
+          facingMode: 'environment' // For phones, prefer main camera
+        }
+      }).then(function(stream) {
+        // Start rendering camera
+        var streamURL = window.URL.createObjectURL(stream);
+        try {
+          camera.srcObject = streamURL;
+        } catch (e) {
+          camera.src = streamURL;
+        }
+
+        // Resolve promise
+        resolve();
+      }).catch(reject);
+    });
+  }
+
+
+  /**
+   * Load model
+   * @async
+   * @return {Promise} Callback
+   */
+  function loadModel() {
+    return new Promise(function(resolve, reject) {
+      tf.loadModel(MODEL_PATH).then(function(m) {
+        model = m;
+        fetch(CLASSES_PATH).then(function(res) {
+          res.json().then(function(json) {
+            modelClasses = json;
+            resolve();
+          });
+        });
+      }).catch(reject);
+    });
+  }
+
+
+  /**
+   * Start predicting
+   */
+  function startPredicting() {
+    predict().then(function(isHotdog) {
+      // Update UI
+      if (isHotdog) console.log('Hotdog found!');
+      updateUI(isHotdog);
+
+      // Schedule next prediction
+      setTimeout(function() {
+        startPredicting();
+      }, PREDICT_INTERVAL);
+    });
+  }
+
+
+  /**
+   * Predict
+   * @async
+   * @return {Promise<Boolean>} Is hotdog
+   */
+  function predict() {
+    return new Promise(function(resolve, reject) {
+      tf.tidy(function() {
+        // Get camera pixels and covert them to a tensor
+        var image = tf.fromPixels(getCameraImage()).toFloat();
+
+        // Change coordinates from [0,255] to [-1,1]
+        var offset = tf.scalar(127.5);
+        image = image.sub(offset).div(offset);
+
+        // Convert linear array to matrix
+        image = image.reshape([1, IMAGE_SIZE, IMAGE_SIZE, 3]);
+
+        // Make a prediction using loaded model
+        var logits = model.predict(image);
+        getTopClasses(logits, MAX_CLASSES).then(function(classes) {
+          console.log('Found classes', classes);
+
+          // Validate whether is Hotdog or Not Hotdog
+          var isHotdog = false;
+          for (var i=0; i<classes.length; i++) {
+            if (classes[i].name.indexOf('hotdog') > -1) {
+              isHotdog = true;
+              break;
+            }
+          }
+          resolve(isHotdog);
+        });
+      });
+    });
+  }
+
+
+  /**
+   * Get top classes
+   * @param  {tf.Tensor}      logits     Logits returned by model
+   * @param  {Integer}        maxClasses Maximum number of classes to return
+   * @return {Promise<Array>}            Top classes
+   */
+  function getTopClasses(logits, maxClasses) {
+    return new Promise(function(resolve, reject) {
+      // Get raw data from logits
+      logits.data().then(function(values) {
+        // Sort data by value, while keeping index
+        var sortedData = [];
+        for (var i=0; i<values.length; i++) {
+          sortedData.push({value: values[i], index: i});
+        }
+        sortedData.sort(function(a, b) {
+          return b.value - a.value;
+        });
+
+        // Get top entries
+        var topValues = new Float32Array(maxClasses);
+        var topIndices = new Int32Array(maxClasses);
+        for (var i=0; i<maxClasses; i++) {
+          topValues[i] = sortedData[i].value;
+          topIndices[i] = sortedData[i].index;
+        }
+
+        // Get top classes name
+        var topClasses = [];
+        for (var i=0; i<topIndices.length; i++) {
+          topClasses.push({
+            name: modelClasses[topIndices[i]],
+            probability: topValues[i]
+          });
+        }
+        resolve(topClasses);
+      });
+    });
+  }
+
+
+  /* INITIALIZE */
+  connectToCamera().then(function() {
+    overlayDialog.innerHTML = 'Cargando modelo...';
+    loadModel().then(function() {
+      overlayDialog.style.display = 'none';
+      startPredicting();
+    });
+  }).catch(function() {
+    overlayDialog.innerHTML = 'Esta app necesita acceder a la cámara de ' +
+      'tu dispositivo para funcionar';
+  });
+
+})();
diff --git a/images/cross.png b/images/cross.png
new file mode 100644
index 0000000..5dcdb7e
Binary files /dev/null and b/images/cross.png differ
diff --git a/images/hotdog.png b/images/hotdog.png
new file mode 100644
index 0000000..5b32998
Binary files /dev/null and b/images/hotdog.png differ
diff --git a/images/icon-144.png b/images/icon-144.png
new file mode 100644
index 0000000..0ab0a9a
Binary files /dev/null and b/images/icon-144.png differ
diff --git a/images/icon-48.png b/images/icon-48.png
new file mode 100644
index 0000000..3c6f4bd
Binary files /dev/null and b/images/icon-48.png differ
diff --git a/images/icon-512.png b/images/icon-512.png
new file mode 100644
index 0000000..5a85c43
Binary files /dev/null and b/images/icon-512.png differ
diff --git a/index.html b/index.html
new file mode 100644
index 0000000..0dcb255
--- /dev/null
+++ b/index.html
@@ -0,0 +1,27 @@
+<!doctype html>
+<html lang="es">
+  <head>
+    <meta charset="utf-8">
+    <meta name="viewport" content="user-scalable=no, initial-scale=1, maximum-scale=1, minimum-scale=1, width=device-width">
+    <title>Hotdog or Not Hotdog</title>
+    <link rel="manifest" href="manifest.json">
+    <meta name="theme-color" content="#21a8c6">
+    <link rel="icon" href="images/hotdog.png">
+    <link rel="stylesheet" href="app.css">
+  </head>
+  <body>
+    <!-- OVERLAY -->
+    <div id="overlay">
+      <div class="result"></div>
+      <div class="dialog">Concede acceso a la cámara para continuar</div>
+    </div>
+
+    <!-- CAMERA UI -->
+    <video id="camera" autoplay></video>
+
+    <!-- SCRIPTS -->
+    <script type="text/javascript" src="worker-installer.js"></script>
+    <script type="text/javascript" src="tensorflow.js"></script>
+    <script type="text/javascript" src="app.js"></script>
+  </body>
+</html>
diff --git a/manifest.json b/manifest.json
new file mode 100644
index 0000000..d81173f
--- /dev/null
+++ b/manifest.json
@@ -0,0 +1,26 @@
+{
+  "name": "Hotdog or Not Hotdog",
+  "short_name": "Not Hotdog",
+  "description": "Detecta perritos calientes mediante el uso de IA",
+  "icons": [
+    {
+      "src": "images/icon-48.png",
+      "type": "image/png",
+      "sizes": "48x48"
+    },
+    {
+      "src": "images/icon-144.png",
+      "type": "image/png",
+      "sizes": "144x144"
+    },
+    {
+      "src": "images/icon-512.png",
+      "type": "image/png",
+      "sizes": "512x512"
+    }
+  ],
+  "start_url": "./",
+  "display": "standalone",
+  "background_color": "#202020",
+  "theme_color": "#21a8c6"
+}
diff --git a/model/group1-shard1of1 b/model/group1-shard1of1
new file mode 100644
index 0000000..bd51711
Binary files /dev/null and b/model/group1-shard1of1 differ
diff --git a/model/group10-shard1of1 b/model/group10-shard1of1
new file mode 100644
index 0000000..be70b8e
Binary files /dev/null and b/model/group10-shard1of1 differ
diff --git a/model/group11-shard1of1 b/model/group11-shard1of1
new file mode 100644
index 0000000..563207c
Binary files /dev/null and b/model/group11-shard1of1 differ
diff --git a/model/group12-shard1of1 b/model/group12-shard1of1
new file mode 100644
index 0000000..51b380b
--- /dev/null
+++ b/model/group12-shard1of1
@@ -0,0 +1,2 @@
+#����?�C>/�=&?�-H���?@����c@��?����֦����@\r�@��R@-;��AB�X��~�M�,��(���{��'D�¢�9�0B��Bu>c��JkC1�
+E�NB���
\ No newline at end of file
diff --git a/model/group13-shard1of1 b/model/group13-shard1of1
new file mode 100644
index 0000000..b09e1f7
--- /dev/null
+++ b/model/group13-shard1of1
@@ -0,0 +1,7 @@
+hT�?)�˿t�#�M7u?�1�><����ѵ�W�?�~�?I�2?f/��҈?��O>�s��鑿�*���i�@z������@�×?
)�*
@ǀh?I�@N�!@QY�@�d@ِ��j��@�*�}����*��;/�q������@�?��:���큾�qU�>��?���@
+�A�s��k��V%���>e�m�>iE��u�6��f�?a��ݛ��'�@��@(=�.~>�
+�ǩI��DŽ�GEm�Q��@Ҷ�?�{����?G7?����@����A'_@P�0?�E�?4���TA+��=���"�@� P=~�>e=ʽFA�?��
�*�e���
+@�D@�*@5ǿo�]@�7�I=E>䅱�?(s��@ڽ�h_�?b�?;3E�Z=�/>0SWAAJ��
+P?
+��?:-�? B�?��?����@���?o�<�,1�?
+��1��.k�R�?k�i&:��D�?��2���%����@�ˑ@�CÿFA�@դ�����'&R��?a�)�X#���࿖��1��?�SY�RV����?	����5j@��3�
\ No newline at end of file
diff --git a/model/group14-shard1of1 b/model/group14-shard1of1
new file mode 100644
index 0000000..e44f706
Binary files /dev/null and b/model/group14-shard1of1 differ
diff --git a/model/group15-shard1of1 b/model/group15-shard1of1
new file mode 100644
index 0000000..4ea0bad
Binary files /dev/null and b/model/group15-shard1of1 differ
diff --git a/model/group16-shard1of1 b/model/group16-shard1of1
new file mode 100644
index 0000000..24e2f17
Binary files /dev/null and b/model/group16-shard1of1 differ
diff --git a/model/group17-shard1of1 b/model/group17-shard1of1
new file mode 100644
index 0000000..44af369
Binary files /dev/null and b/model/group17-shard1of1 differ
diff --git a/model/group18-shard1of1 b/model/group18-shard1of1
new file mode 100644
index 0000000..c5154bc
--- /dev/null
+++ b/model/group18-shard1of1
@@ -0,0 +1,2 @@
+&�b?��b?s�N?m��?<W?8�F??�<?dň?��]?@?:��?��V?�s?(%�?�{@O$L?Z7?3!d?�/O?Y;?NQ?�q?�%3?5�>?Њ|?�.P?r/]?_Q=?C/T?��T?/��>}<?��?���?�˃@�,��
+��?>��?3�@�@�
�@�h?u��@Ll�?�M�@�@$)��0x?	�@v�E@�?�L{?�^�?�-�@��@�$�@�;G=�Wt?���?^Z�?oL�?�R@���>���?���qw&A���@]�5@�,*A7uA�.����A���]A�M1@24Ak��� A���@K�AM��A�0B:�A��A�%�@n�+����A9>���W�>���@�|�B���@��@�@Ss�?��A���B�6�BjWB,B�A���B8@�B�4�A��B�hB��B��B�V�B�`
C���B_oVB->_B(��Cb�C�7�B|BsC4��A��@���A��0B��B��#B�^C��A��!B7��>#&A�B
B
\ No newline at end of file
diff --git a/model/group19-shard1of1 b/model/group19-shard1of1
new file mode 100644
index 0000000..775d03b
Binary files /dev/null and b/model/group19-shard1of1 differ
diff --git a/model/group2-shard1of1 b/model/group2-shard1of1
new file mode 100644
index 0000000..f30b366
--- /dev/null
+++ b/model/group2-shard1of1
@@ -0,0 +1,4 @@
+��>
+�T�
+̆?�0�?@���#GQA���>�u�?��@��=@s�>8Ǿ�V&@v�@�uF@:���� B?ȭ���[Q9y��<G�>e��<_+8=}<�*���@�$@�U5
+��F8�@\3�=(@D5
\ No newline at end of file
diff --git a/model/group20-shard1of1 b/model/group20-shard1of1
new file mode 100644
index 0000000..4b2809c
Binary files /dev/null and b/model/group20-shard1of1 differ
diff --git a/model/group21-shard1of1 b/model/group21-shard1of1
new file mode 100644
index 0000000..8e708b0
Binary files /dev/null and b/model/group21-shard1of1 differ
diff --git a/model/group22-shard1of1 b/model/group22-shard1of1
new file mode 100644
index 0000000..0703fc1
Binary files /dev/null and b/model/group22-shard1of1 differ
diff --git a/model/group23-shard1of1 b/model/group23-shard1of1
new file mode 100644
index 0000000..7f5b536
Binary files /dev/null and b/model/group23-shard1of1 differ
diff --git a/model/group24-shard1of1 b/model/group24-shard1of1
new file mode 100644
index 0000000..27a0a62
Binary files /dev/null and b/model/group24-shard1of1 differ
diff --git a/model/group25-shard1of1 b/model/group25-shard1of1
new file mode 100644
index 0000000..5b8e707
Binary files /dev/null and b/model/group25-shard1of1 differ
diff --git a/model/group26-shard1of1 b/model/group26-shard1of1
new file mode 100644
index 0000000..6cb4195
Binary files /dev/null and b/model/group26-shard1of1 differ
diff --git a/model/group27-shard1of1 b/model/group27-shard1of1
new file mode 100644
index 0000000..ca8c664
Binary files /dev/null and b/model/group27-shard1of1 differ
diff --git a/model/group28-shard1of1 b/model/group28-shard1of1
new file mode 100644
index 0000000..daaad27
Binary files /dev/null and b/model/group28-shard1of1 differ
diff --git a/model/group29-shard1of1 b/model/group29-shard1of1
new file mode 100644
index 0000000..7b6183e
Binary files /dev/null and b/model/group29-shard1of1 differ
diff --git a/model/group3-shard1of1 b/model/group3-shard1of1
new file mode 100644
index 0000000..53b16cf
Binary files /dev/null and b/model/group3-shard1of1 differ
diff --git a/model/group30-shard1of1 b/model/group30-shard1of1
new file mode 100644
index 0000000..b2e2a50
Binary files /dev/null and b/model/group30-shard1of1 differ
diff --git a/model/group31-shard1of1 b/model/group31-shard1of1
new file mode 100644
index 0000000..16d6bfe
Binary files /dev/null and b/model/group31-shard1of1 differ
diff --git a/model/group32-shard1of1 b/model/group32-shard1of1
new file mode 100644
index 0000000..d421186
Binary files /dev/null and b/model/group32-shard1of1 differ
diff --git a/model/group33-shard1of1 b/model/group33-shard1of1
new file mode 100644
index 0000000..c36c071
Binary files /dev/null and b/model/group33-shard1of1 differ
diff --git a/model/group34-shard1of1 b/model/group34-shard1of1
new file mode 100644
index 0000000..cb85b68
Binary files /dev/null and b/model/group34-shard1of1 differ
diff --git a/model/group35-shard1of1 b/model/group35-shard1of1
new file mode 100644
index 0000000..eb69357
Binary files /dev/null and b/model/group35-shard1of1 differ
diff --git a/model/group36-shard1of1 b/model/group36-shard1of1
new file mode 100644
index 0000000..817344e
Binary files /dev/null and b/model/group36-shard1of1 differ
diff --git a/model/group37-shard1of1 b/model/group37-shard1of1
new file mode 100644
index 0000000..ecd1d91
Binary files /dev/null and b/model/group37-shard1of1 differ
diff --git a/model/group38-shard1of1 b/model/group38-shard1of1
new file mode 100644
index 0000000..2e0262b
Binary files /dev/null and b/model/group38-shard1of1 differ
diff --git a/model/group39-shard1of1 b/model/group39-shard1of1
new file mode 100644
index 0000000..e347328
Binary files /dev/null and b/model/group39-shard1of1 differ
diff --git a/model/group4-shard1of1 b/model/group4-shard1of1
new file mode 100644
index 0000000..f89910b
Binary files /dev/null and b/model/group4-shard1of1 differ
diff --git a/model/group40-shard1of1 b/model/group40-shard1of1
new file mode 100644
index 0000000..8cd80f1
Binary files /dev/null and b/model/group40-shard1of1 differ
diff --git a/model/group41-shard1of1 b/model/group41-shard1of1
new file mode 100644
index 0000000..4eb76b2
--- /dev/null
+++ b/model/group41-shard1of1
@@ -0,0 +1,5 @@
+"�>̥�!?⿄?�D�@�0�@�9�>�d�?]e�?��b?P�M?�NW?~��?�u?�H.>IN�?��?,�Y?:�r>�?}�2?~��?��>dI�?��S?X��@�5?��(?/�:?���>%��?1�.?�q@i���,`V@r�?��=@4j@�:�@^��>1��@��*@��%@�C4@�o�?�G$@�{����=��@|�(@.�m�?�Fh@��9@�5�?���?�
+�?�$b@T8Q@/Y?5U�?��2@���?t8@�v�>|z��@
+D�Փ@��Y�Kw@H����))���&�/�3��[v�6D��
�e@p\h��qϾJ6�=1_?H��*���@��%�TJ�@�v!AUk����*
+�?�r���=ֿ�
+�y�?2��?Qn�@J#?٭g<�?z��?\��>��>�ip?2@?�e�?S�D?�E�?���?wA??&��?U�suxi@�~�?�[�?�A\[��>�:-@4N�>/��?�	?��
?no7?G1�?1c?��>(A7?��
?BE@@
\ No newline at end of file
diff --git a/model/group42-shard1of1 b/model/group42-shard1of1
new file mode 100644
index 0000000..87e3894
Binary files /dev/null and b/model/group42-shard1of1 differ
diff --git a/model/group43-shard1of1 b/model/group43-shard1of1
new file mode 100644
index 0000000..2b8a213
Binary files /dev/null and b/model/group43-shard1of1 differ
diff --git a/model/group44-shard1of1 b/model/group44-shard1of1
new file mode 100644
index 0000000..d022176
Binary files /dev/null and b/model/group44-shard1of1 differ
diff --git a/model/group45-shard1of1 b/model/group45-shard1of1
new file mode 100644
index 0000000..dfe4da8
Binary files /dev/null and b/model/group45-shard1of1 differ
diff --git a/model/group46-shard1of1 b/model/group46-shard1of1
new file mode 100644
index 0000000..1640dd8
Binary files /dev/null and b/model/group46-shard1of1 differ
diff --git a/model/group47-shard1of1 b/model/group47-shard1of1
new file mode 100644
index 0000000..a03d04d
Binary files /dev/null and b/model/group47-shard1of1 differ
diff --git a/model/group48-shard1of1 b/model/group48-shard1of1
new file mode 100644
index 0000000..15b3ae9
Binary files /dev/null and b/model/group48-shard1of1 differ
diff --git a/model/group49-shard1of1 b/model/group49-shard1of1
new file mode 100644
index 0000000..43cd606
Binary files /dev/null and b/model/group49-shard1of1 differ
diff --git a/model/group5-shard1of1 b/model/group5-shard1of1
new file mode 100644
index 0000000..bd227b0
Binary files /dev/null and b/model/group5-shard1of1 differ
diff --git a/model/group50-shard1of1 b/model/group50-shard1of1
new file mode 100644
index 0000000..fe1b4a6
Binary files /dev/null and b/model/group50-shard1of1 differ
diff --git a/model/group51-shard1of1 b/model/group51-shard1of1
new file mode 100644
index 0000000..cfe87a6
Binary files /dev/null and b/model/group51-shard1of1 differ
diff --git a/model/group52-shard1of1 b/model/group52-shard1of1
new file mode 100644
index 0000000..5d34e33
Binary files /dev/null and b/model/group52-shard1of1 differ
diff --git a/model/group53-shard1of1 b/model/group53-shard1of1
new file mode 100644
index 0000000..36abaef
Binary files /dev/null and b/model/group53-shard1of1 differ
diff --git a/model/group54-shard1of1 b/model/group54-shard1of1
new file mode 100644
index 0000000..c496eb1
Binary files /dev/null and b/model/group54-shard1of1 differ
diff --git a/model/group55-shard1of1 b/model/group55-shard1of1
new file mode 100644
index 0000000..b9d8357
Binary files /dev/null and b/model/group55-shard1of1 differ
diff --git a/model/group6-shard1of1 b/model/group6-shard1of1
new file mode 100644
index 0000000..4231264
Binary files /dev/null and b/model/group6-shard1of1 differ
diff --git a/model/group7-shard1of1 b/model/group7-shard1of1
new file mode 100644
index 0000000..763abe2
Binary files /dev/null and b/model/group7-shard1of1 differ
diff --git a/model/group8-shard1of1 b/model/group8-shard1of1
new file mode 100644
index 0000000..52bc94d
Binary files /dev/null and b/model/group8-shard1of1 differ
diff --git a/model/group9-shard1of1 b/model/group9-shard1of1
new file mode 100644
index 0000000..727443f
Binary files /dev/null and b/model/group9-shard1of1 differ
diff --git a/model/mobilenet.classes.json b/model/mobilenet.classes.json
new file mode 100644
index 0000000..8bdcfc8
--- /dev/null
+++ b/model/mobilenet.classes.json
@@ -0,0 +1 @@
+{"0":"tench, Tinca tinca","1":"goldfish, Carassius auratus","2":"great white shark, white shark, man-eater, man-eating shark, Carcharodon carcharias","3":"tiger shark, Galeocerdo cuvieri","4":"hammerhead, hammerhead shark","5":"electric ray, crampfish, numbfish, torpedo","6":"stingray","7":"cock","8":"hen","9":"ostrich, Struthio camelus","10":"brambling, Fringilla montifringilla","11":"goldfinch, Carduelis carduelis","12":"house finch, linnet, Carpodacus mexicanus","13":"junco, snowbird","14":"indigo bunting, indigo finch, indigo bird, Passerina cyanea","15":"robin, American robin, Turdus migratorius","16":"bulbul","17":"jay","18":"magpie","19":"chickadee","20":"water ouzel, dipper","21":"kite","22":"bald eagle, American eagle, Haliaeetus leucocephalus","23":"vulture","24":"great grey owl, great gray owl, Strix nebulosa","25":"European fire salamander, Salamandra salamandra","26":"common newt, Triturus vulgaris","27":"eft","28":"spotted salamander, Ambystoma maculatum","29":"axolotl, mud puppy, Ambystoma mexicanum","30":"bullfrog, Rana catesbeiana","31":"tree frog, tree-frog","32":"tailed frog, bell toad, ribbed toad, tailed toad, Ascaphus trui","33":"loggerhead, loggerhead turtle, Caretta caretta","34":"leatherback turtle, leatherback, leathery turtle, Dermochelys coriacea","35":"mud turtle","36":"terrapin","37":"box turtle, box tortoise","38":"banded gecko","39":"common iguana, iguana, Iguana iguana","40":"American chameleon, anole, Anolis carolinensis","41":"whiptail, whiptail lizard","42":"agama","43":"frilled lizard, Chlamydosaurus kingi","44":"alligator lizard","45":"Gila monster, Heloderma suspectum","46":"green lizard, Lacerta viridis","47":"African chameleon, Chamaeleo chamaeleon","48":"Komodo dragon, Komodo lizard, dragon lizard, giant lizard, Varanus komodoensis","49":"African crocodile, Nile crocodile, Crocodylus niloticus","50":"American alligator, Alligator mississipiensis","51":"triceratops","52":"thunder snake, worm snake, Carphophis amoenus","53":"ringneck snake, ring-necked snake, ring snake","54":"hognose snake, puff adder, sand viper","55":"green snake, grass snake","56":"king snake, kingsnake","57":"garter snake, grass snake","58":"water snake","59":"vine snake","60":"night snake, Hypsiglena torquata","61":"boa constrictor, Constrictor constrictor","62":"rock python, rock snake, Python sebae","63":"Indian cobra, Naja naja","64":"green mamba","65":"sea snake","66":"horned viper, cerastes, sand viper, horned asp, Cerastes cornutus","67":"diamondback, diamondback rattlesnake, Crotalus adamanteus","68":"sidewinder, horned rattlesnake, Crotalus cerastes","69":"trilobite","70":"harvestman, daddy longlegs, Phalangium opilio","71":"scorpion","72":"black and gold garden spider, Argiope aurantia","73":"barn spider, Araneus cavaticus","74":"garden spider, Aranea diademata","75":"black widow, Latrodectus mactans","76":"tarantula","77":"wolf spider, hunting spider","78":"tick","79":"centipede","80":"black grouse","81":"ptarmigan","82":"ruffed grouse, partridge, Bonasa umbellus","83":"prairie chicken, prairie grouse, prairie fowl","84":"peacock","85":"quail","86":"partridge","87":"African grey, African gray, Psittacus erithacus","88":"macaw","89":"sulphur-crested cockatoo, Kakatoe galerita, Cacatua galerita","90":"lorikeet","91":"coucal","92":"bee eater","93":"hornbill","94":"hummingbird","95":"jacamar","96":"toucan","97":"drake","98":"red-breasted merganser, Mergus serrator","99":"goose","100":"black swan, Cygnus atratus","101":"tusker","102":"echidna, spiny anteater, anteater","103":"platypus, duckbill, duckbilled platypus, duck-billed platypus, Ornithorhynchus anatinus","104":"wallaby, brush kangaroo","105":"koala, koala bear, kangaroo bear, native bear, Phascolarctos cinereus","106":"wombat","107":"jelly fish","108":"sea anemone, anemone","109":"brain coral","110":"flatworm, platyhelminth","111":"nematode, nematode worm, roundworm","112":"conch","113":"snail","114":"slug","115":"sea slug, nudibranch","116":"chiton, coat-of-mail shell, sea cradle, polyplacophore","117":"chambered nautilus, pearly nautilus, nautilus","118":"Dungeness crab, Cancer magister","119":"rock crab, Cancer irroratus","120":"fiddler crab","121":"king crab, Alaska crab, Alaskan king crab, Alaska king crab, Paralithodes camtschatica","122":"American lobster, Northern lobster, Maine lobster, Homarus americanus","123":"spiny lobster, langouste, rock lobster, crawfish, crayfish, sea crawfish","124":"crayfish, crawfish, crawdad, crawdaddy","125":"hermit crab","126":"isopod","127":"white stork, Ciconia ciconia","128":"black stork, Ciconia nigra","129":"spoonbill","130":"flamingo","131":"little blue heron, Egretta caerulea","132":"American egret, great white heron, Egretta albus","133":"bittern","134":"crane","135":"limpkin, Aramus pictus","136":"European gallinule, Porphyrio porphyrio","137":"American coot, marsh hen, mud hen, water hen, Fulica americana","138":"bustard","139":"ruddy turnstone, Arenaria interpres","140":"red-backed sandpiper, dunlin, Erolia alpina","141":"redshank, Tringa totanus","142":"dowitcher","143":"oystercatcher, oyster catcher","144":"pelican","145":"king penguin, Aptenodytes patagonica","146":"albatross, mollymawk","147":"grey whale, gray whale, devilfish, Eschrichtius gibbosus, Eschrichtius robustus","148":"killer whale, killer, orca, grampus, sea wolf, Orcinus orca","149":"dugong, Dugong dugon","150":"sea lion","151":"Chihuahua","152":"Japanese spaniel","153":"Maltese dog, Maltese terrier, Maltese","154":"Pekinese, Pekingese, Peke","155":"Shih-Tzu","156":"Blenheim spaniel","157":"papillon","158":"toy terrier","159":"Rhodesian ridgeback","160":"Afghan hound, Afghan","161":"basset, basset hound","162":"beagle","163":"bloodhound, sleuthhound","164":"bluetick","165":"black-and-tan coonhound","166":"Walker hound, Walker foxhound","167":"English foxhound","168":"redbone","169":"borzoi, Russian wolfhound","170":"Irish wolfhound","171":"Italian greyhound","172":"whippet","173":"Ibizan hound, Ibizan Podenco","174":"Norwegian elkhound, elkhound","175":"otterhound, otter hound","176":"Saluki, gazelle hound","177":"Scottish deerhound, deerhound","178":"Weimaraner","179":"Staffordshire bullterrier, Staffordshire bull terrier","180":"American Staffordshire terrier, Staffordshire terrier, American pit bull terrier, pit bull terrier","181":"Bedlington terrier","182":"Border terrier","183":"Kerry blue terrier","184":"Irish terrier","185":"Norfolk terrier","186":"Norwich terrier","187":"Yorkshire terrier","188":"wire-haired fox terrier","189":"Lakeland terrier","190":"Sealyham terrier, Sealyham","191":"Airedale, Airedale terrier","192":"cairn, cairn terrier","193":"Australian terrier","194":"Dandie Dinmont, Dandie Dinmont terrier","195":"Boston bull, Boston terrier","196":"miniature schnauzer","197":"giant schnauzer","198":"standard schnauzer","199":"Scotch terrier, Scottish terrier, Scottie","200":"Tibetan terrier, chrysanthemum dog","201":"silky terrier, Sydney silky","202":"soft-coated wheaten terrier","203":"West Highland white terrier","204":"Lhasa, Lhasa apso","205":"flat-coated retriever","206":"curly-coated retriever","207":"golden retriever","208":"Labrador retriever","209":"Chesapeake Bay retriever","210":"German short-haired pointer","211":"vizsla, Hungarian pointer","212":"English setter","213":"Irish setter, red setter","214":"Gordon setter","215":"Brittany spaniel","216":"clumber, clumber spaniel","217":"English springer, English springer spaniel","218":"Welsh springer spaniel","219":"cocker spaniel, English cocker spaniel, cocker","220":"Sussex spaniel","221":"Irish water spaniel","222":"kuvasz","223":"schipperke","224":"groenendael","225":"malinois","226":"briard","227":"kelpie","228":"komondor","229":"Old English sheepdog, bobtail","230":"Shetland sheepdog, Shetland sheep dog, Shetland","231":"collie","232":"Border collie","233":"Bouvier des Flandres, Bouviers des Flandres","234":"Rottweiler","235":"German shepherd, German shepherd dog, German police dog, alsatian","236":"Doberman, Doberman pinscher","237":"miniature pinscher","238":"Greater Swiss Mountain dog","239":"Bernese mountain dog","240":"Appenzeller","241":"EntleBucher","242":"boxer","243":"bull mastiff","244":"Tibetan mastiff","245":"French bulldog","246":"Great Dane","247":"Saint Bernard, St Bernard","248":"Eskimo dog, husky","249":"malamute, malemute, Alaskan malamute","250":"Siberian husky","251":"dalmatian, coach dog, carriage dog","252":"affenpinscher, monkey pinscher, monkey dog","253":"basenji","254":"pug, pug-dog","255":"Leonberg","256":"Newfoundland, Newfoundland dog","257":"Great Pyrenees","258":"Samoyed, Samoyede","259":"Pomeranian","260":"chow, chow chow","261":"keeshond","262":"Brabancon griffon","263":"Pembroke, Pembroke Welsh corgi","264":"Cardigan, Cardigan Welsh corgi","265":"toy poodle","266":"miniature poodle","267":"standard poodle","268":"Mexican hairless","269":"timber wolf, grey wolf, gray wolf, Canis lupus","270":"white wolf, Arctic wolf, Canis lupus tundrarum","271":"red wolf, maned wolf, Canis rufus, Canis niger","272":"coyote, prairie wolf, brush wolf, Canis latrans","273":"dingo, warrigal, warragal, Canis dingo","274":"dhole, Cuon alpinus","275":"African hunting dog, hyena dog, Cape hunting dog, Lycaon pictus","276":"hyena, hyaena","277":"red fox, Vulpes vulpes","278":"kit fox, Vulpes macrotis","279":"Arctic fox, white fox, Alopex lagopus","280":"grey fox, gray fox, Urocyon cinereoargenteus","281":"tabby, tabby cat","282":"tiger cat","283":"Persian cat","284":"Siamese cat, Siamese","285":"Egyptian cat","286":"cougar, puma, catamount, mountain lion, painter, panther, Felis concolor","287":"lynx, catamount","288":"leopard, Panthera pardus","289":"snow leopard, ounce, Panthera uncia","290":"jaguar, panther, Panthera onca, Felis onca","291":"lion, king of beasts, Panthera leo","292":"tiger, Panthera tigris","293":"cheetah, chetah, Acinonyx jubatus","294":"brown bear, bruin, Ursus arctos","295":"American black bear, black bear, Ursus americanus, Euarctos americanus","296":"ice bear, polar bear, Ursus Maritimus, Thalarctos maritimus","297":"sloth bear, Melursus ursinus, Ursus ursinus","298":"mongoose","299":"meerkat, mierkat","300":"tiger beetle","301":"ladybug, ladybeetle, lady beetle, ladybird, ladybird beetle","302":"ground beetle, carabid beetle","303":"long-horned beetle, longicorn, longicorn beetle","304":"leaf beetle, chrysomelid","305":"dung beetle","306":"rhinoceros beetle","307":"weevil","308":"fly","309":"bee","310":"ant, emmet, pismire","311":"grasshopper, hopper","312":"cricket","313":"walking stick, walkingstick, stick insect","314":"cockroach, roach","315":"mantis, mantid","316":"cicada, cicala","317":"leafhopper","318":"lacewing, lacewing fly","319":"dragonfly, darning needle, devil's darning needle, sewing needle, snake feeder, snake doctor, mosquito hawk, skeeter hawk","320":"damselfly","321":"admiral","322":"ringlet, ringlet butterfly","323":"monarch, monarch butterfly, milkweed butterfly, Danaus plexippus","324":"cabbage butterfly","325":"sulphur butterfly, sulfur butterfly","326":"lycaenid, lycaenid butterfly","327":"starfish, sea star","328":"sea urchin","329":"sea cucumber, holothurian","330":"wood rabbit, cottontail, cottontail rabbit","331":"hare","332":"Angora, Angora rabbit","333":"hamster","334":"porcupine, hedgehog","335":"fox squirrel, eastern fox squirrel, Sciurus niger","336":"marmot","337":"beaver","338":"guinea pig, Cavia cobaya","339":"sorrel","340":"zebra","341":"hog, pig, grunter, squealer, Sus scrofa","342":"wild boar, boar, Sus scrofa","343":"warthog","344":"hippopotamus, hippo, river horse, Hippopotamus amphibius","345":"ox","346":"water buffalo, water ox, Asiatic buffalo, Bubalus bubalis","347":"bison","348":"ram, tup","349":"bighorn, bighorn sheep, cimarron, Rocky Mountain bighorn, Rocky Mountain sheep, Ovis canadensis","350":"ibex, Capra ibex","351":"hartebeest","352":"impala, Aepyceros melampus","353":"gazelle","354":"Arabian camel, dromedary, Camelus dromedarius","355":"llama","356":"weasel","357":"mink","358":"polecat, fitch, foulmart, foumart, Mustela putorius","359":"black-footed ferret, ferret, Mustela nigripes","360":"otter","361":"skunk, polecat, wood pussy","362":"badger","363":"armadillo","364":"three-toed sloth, ai, Bradypus tridactylus","365":"orangutan, orang, orangutang, Pongo pygmaeus","366":"gorilla, Gorilla gorilla","367":"chimpanzee, chimp, Pan troglodytes","368":"gibbon, Hylobates lar","369":"siamang, Hylobates syndactylus, Symphalangus syndactylus","370":"guenon, guenon monkey","371":"patas, hussar monkey, Erythrocebus patas","372":"baboon","373":"macaque","374":"langur","375":"colobus, colobus monkey","376":"proboscis monkey, Nasalis larvatus","377":"marmoset","378":"capuchin, ringtail, Cebus capucinus","379":"howler monkey, howler","380":"titi, titi monkey","381":"spider monkey, Ateles geoffroyi","382":"squirrel monkey, Saimiri sciureus","383":"Madagascar cat, ring-tailed lemur, Lemur catta","384":"indri, indris, Indri indri, Indri brevicaudatus","385":"Indian elephant, Elephas maximus","386":"African elephant, Loxodonta africana","387":"lesser panda, red panda, panda, bear cat, cat bear, Ailurus fulgens","388":"giant panda, panda, panda bear, coon bear, Ailuropoda melanoleuca","389":"barracouta, snoek","390":"eel","391":"coho, cohoe, coho salmon, blue jack, silver salmon, Oncorhynchus kisutch","392":"rock beauty, Holocanthus tricolor","393":"anemone fish","394":"sturgeon","395":"gar, garfish, garpike, billfish, Lepisosteus osseus","396":"lionfish","397":"puffer, pufferfish, blowfish, globefish","398":"abacus","399":"abaya","400":"academic gown, academic robe, judge's robe","401":"accordion, piano accordion, squeeze box","402":"acoustic guitar","403":"aircraft carrier, carrier, flattop, attack aircraft carrier","404":"airliner","405":"airship, dirigible","406":"altar","407":"ambulance","408":"amphibian, amphibious vehicle","409":"analog clock","410":"apiary, bee house","411":"apron","412":"ashcan, trash can, garbage can, wastebin, ash bin, ash-bin, ashbin, dustbin, trash barrel, trash bin","413":"assault rifle, assault gun","414":"backpack, back pack, knapsack, packsack, rucksack, haversack","415":"bakery, bakeshop, bakehouse","416":"balance beam, beam","417":"balloon","418":"ballpoint, ballpoint pen, ballpen, Biro","419":"Band Aid","420":"banjo","421":"bannister, banister, balustrade, balusters, handrail","422":"barbell","423":"barber chair","424":"barbershop","425":"barn","426":"barometer","427":"barrel, cask","428":"barrow, garden cart, lawn cart, wheelbarrow","429":"baseball","430":"basketball","431":"bassinet","432":"bassoon","433":"bathing cap, swimming cap","434":"bath towel","435":"bathtub, bathing tub, bath, tub","436":"beach wagon, station wagon, wagon, estate car, beach waggon, station waggon, waggon","437":"beacon, lighthouse, beacon light, pharos","438":"beaker","439":"bearskin, busby, shako","440":"beer bottle","441":"beer glass","442":"bell cote, bell cot","443":"bib","444":"bicycle-built-for-two, tandem bicycle, tandem","445":"bikini, two-piece","446":"binder, ring-binder","447":"binoculars, field glasses, opera glasses","448":"birdhouse","449":"boathouse","450":"bobsled, bobsleigh, bob","451":"bolo tie, bolo, bola tie, bola","452":"bonnet, poke bonnet","453":"bookcase","454":"bookshop, bookstore, bookstall","455":"bottlecap","456":"bow","457":"bow tie, bow-tie, bowtie","458":"brass, memorial tablet, plaque","459":"brassiere, bra, bandeau","460":"breakwater, groin, groyne, mole, bulwark, seawall, jetty","461":"breastplate, aegis, egis","462":"broom","463":"bucket, pail","464":"buckle","465":"bulletproof vest","466":"bullet train, bullet","467":"butcher shop, meat market","468":"cab, hack, taxi, taxicab","469":"caldron, cauldron","470":"candle, taper, wax light","471":"cannon","472":"canoe","473":"can opener, tin opener","474":"cardigan","475":"car mirror","476":"carousel, carrousel, merry-go-round, roundabout, whirligig","477":"carpenter's kit, tool kit","478":"carton","479":"car wheel","480":"cash machine, cash dispenser, automated teller machine, automatic teller machine, automated teller, automatic teller, ATM","481":"cassette","482":"cassette player","483":"castle","484":"catamaran","485":"CD player","486":"cello, violoncello","487":"cellular telephone, cellular phone, cellphone, cell, mobile phone","488":"chain","489":"chainlink fence","490":"chain mail, ring mail, mail, chain armor, chain armour, ring armor, ring armour","491":"chain saw, chainsaw","492":"chest","493":"chiffonier, commode","494":"chime, bell, gong","495":"china cabinet, china closet","496":"Christmas stocking","497":"church, church building","498":"cinema, movie theater, movie theatre, movie house, picture palace","499":"cleaver, meat cleaver, chopper","500":"cliff dwelling","501":"cloak","502":"clog, geta, patten, sabot","503":"cocktail shaker","504":"coffee mug","505":"coffeepot","506":"coil, spiral, volute, whorl, helix","507":"combination lock","508":"computer keyboard, keypad","509":"confectionery, confectionary, candy store","510":"container ship, containership, container vessel","511":"convertible","512":"corkscrew, bottle screw","513":"cornet, horn, trumpet, trump","514":"cowboy boot","515":"cowboy hat, ten-gallon hat","516":"cradle","517":"crane","518":"crash helmet","519":"crate","520":"crib, cot","521":"Crock Pot","522":"croquet ball","523":"crutch","524":"cuirass","525":"dam, dike, dyke","526":"desk","527":"desktop computer","528":"dial telephone, dial phone","529":"diaper, nappy, napkin","530":"digital clock","531":"digital watch","532":"dining table, board","533":"dishrag, dishcloth","534":"dishwasher, dish washer, dishwashing machine","535":"disk brake, disc brake","536":"dock, dockage, docking facility","537":"dogsled, dog sled, dog sleigh","538":"dome","539":"doormat, welcome mat","540":"drilling platform, offshore rig","541":"drum, membranophone, tympan","542":"drumstick","543":"dumbbell","544":"Dutch oven","545":"electric fan, blower","546":"electric guitar","547":"electric locomotive","548":"entertainment center","549":"envelope","550":"espresso maker","551":"face powder","552":"feather boa, boa","553":"file, file cabinet, filing cabinet","554":"fireboat","555":"fire engine, fire truck","556":"fire screen, fireguard","557":"flagpole, flagstaff","558":"flute, transverse flute","559":"folding chair","560":"football helmet","561":"forklift","562":"fountain","563":"fountain pen","564":"four-poster","565":"freight car","566":"French horn, horn","567":"frying pan, frypan, skillet","568":"fur coat","569":"garbage truck, dustcart","570":"gasmask, respirator, gas helmet","571":"gas pump, gasoline pump, petrol pump, island dispenser","572":"goblet","573":"go-kart","574":"golf ball","575":"golfcart, golf cart","576":"gondola","577":"gong, tam-tam","578":"gown","579":"grand piano, grand","580":"greenhouse, nursery, glasshouse","581":"grille, radiator grille","582":"grocery store, grocery, food market, market","583":"guillotine","584":"hair slide","585":"hair spray","586":"half track","587":"hammer","588":"hamper","589":"hand blower, blow dryer, blow drier, hair dryer, hair drier","590":"hand-held computer, hand-held microcomputer","591":"handkerchief, hankie, hanky, hankey","592":"hard disc, hard disk, fixed disk","593":"harmonica, mouth organ, harp, mouth harp","594":"harp","595":"harvester, reaper","596":"hatchet","597":"holster","598":"home theater, home theatre","599":"honeycomb","600":"hook, claw","601":"hoopskirt, crinoline","602":"horizontal bar, high bar","603":"horse cart, horse-cart","604":"hourglass","605":"iPod","606":"iron, smoothing iron","607":"jack-o'-lantern","608":"jean, blue jean, denim","609":"jeep, landrover","610":"jersey, T-shirt, tee shirt","611":"jigsaw puzzle","612":"jinrikisha, ricksha, rickshaw","613":"joystick","614":"kimono","615":"knee pad","616":"knot","617":"lab coat, laboratory coat","618":"ladle","619":"lampshade, lamp shade","620":"laptop, laptop computer","621":"lawn mower, mower","622":"lens cap, lens cover","623":"letter opener, paper knife, paperknife","624":"library","625":"lifeboat","626":"lighter, light, igniter, ignitor","627":"limousine, limo","628":"liner, ocean liner","629":"lipstick, lip rouge","630":"Loafer","631":"lotion","632":"loudspeaker, speaker, speaker unit, loudspeaker system, speaker system","633":"loupe, jeweler's loupe","634":"lumbermill, sawmill","635":"magnetic compass","636":"mailbag, postbag","637":"mailbox, letter box","638":"maillot","639":"maillot, tank suit","640":"manhole cover","641":"maraca","642":"marimba, xylophone","643":"mask","644":"matchstick","645":"maypole","646":"maze, labyrinth","647":"measuring cup","648":"medicine chest, medicine cabinet","649":"megalith, megalithic structure","650":"microphone, mike","651":"microwave, microwave oven","652":"military uniform","653":"milk can","654":"minibus","655":"miniskirt, mini","656":"minivan","657":"missile","658":"mitten","659":"mixing bowl","660":"mobile home, manufactured home","661":"Model T","662":"modem","663":"monastery","664":"monitor","665":"moped","666":"mortar","667":"mortarboard","668":"mosque","669":"mosquito net","670":"motor scooter, scooter","671":"mountain bike, all-terrain bike, off-roader","672":"mountain tent","673":"mouse, computer mouse","674":"mousetrap","675":"moving van","676":"muzzle","677":"nail","678":"neck brace","679":"necklace","680":"nipple","681":"notebook, notebook computer","682":"obelisk","683":"oboe, hautboy, hautbois","684":"ocarina, sweet potato","685":"odometer, hodometer, mileometer, milometer","686":"oil filter","687":"organ, pipe organ","688":"oscilloscope, scope, cathode-ray oscilloscope, CRO","689":"overskirt","690":"oxcart","691":"oxygen mask","692":"packet","693":"paddle, boat paddle","694":"paddlewheel, paddle wheel","695":"padlock","696":"paintbrush","697":"pajama, pyjama, pj's, jammies","698":"palace","699":"panpipe, pandean pipe, syrinx","700":"paper towel","701":"parachute, chute","702":"parallel bars, bars","703":"park bench","704":"parking meter","705":"passenger car, coach, carriage","706":"patio, terrace","707":"pay-phone, pay-station","708":"pedestal, plinth, footstall","709":"pencil box, pencil case","710":"pencil sharpener","711":"perfume, essence","712":"Petri dish","713":"photocopier","714":"pick, plectrum, plectron","715":"pickelhaube","716":"picket fence, paling","717":"pickup, pickup truck","718":"pier","719":"piggy bank, penny bank","720":"pill bottle","721":"pillow","722":"ping-pong ball","723":"pinwheel","724":"pirate, pirate ship","725":"pitcher, ewer","726":"plane, carpenter's plane, woodworking plane","727":"planetarium","728":"plastic bag","729":"plate rack","730":"plow, plough","731":"plunger, plumber's helper","732":"Polaroid camera, Polaroid Land camera","733":"pole","734":"police van, police wagon, paddy wagon, patrol wagon, wagon, black Maria","735":"poncho","736":"pool table, billiard table, snooker table","737":"pop bottle, soda bottle","738":"pot, flowerpot","739":"potter's wheel","740":"power drill","741":"prayer rug, prayer mat","742":"printer","743":"prison, prison house","744":"projectile, missile","745":"projector","746":"puck, hockey puck","747":"punching bag, punch bag, punching ball, punchball","748":"purse","749":"quill, quill pen","750":"quilt, comforter, comfort, puff","751":"racer, race car, racing car","752":"racket, racquet","753":"radiator","754":"radio, wireless","755":"radio telescope, radio reflector","756":"rain barrel","757":"recreational vehicle, RV, R.V.","758":"reel","759":"reflex camera","760":"refrigerator, icebox","761":"remote control, remote","762":"restaurant, eating house, eating place, eatery","763":"revolver, six-gun, six-shooter","764":"rifle","765":"rocking chair, rocker","766":"rotisserie","767":"rubber eraser, rubber, pencil eraser","768":"rugby ball","769":"rule, ruler","770":"running shoe","771":"safe","772":"safety pin","773":"saltshaker, salt shaker","774":"sandal","775":"sarong","776":"sax, saxophone","777":"scabbard","778":"scale, weighing machine","779":"school bus","780":"schooner","781":"scoreboard","782":"screen, CRT screen","783":"screw","784":"screwdriver","785":"seat belt, seatbelt","786":"sewing machine","787":"shield, buckler","788":"shoe shop, shoe-shop, shoe store","789":"shoji","790":"shopping basket","791":"shopping cart","792":"shovel","793":"shower cap","794":"shower curtain","795":"ski","796":"ski mask","797":"sleeping bag","798":"slide rule, slipstick","799":"sliding door","800":"slot, one-armed bandit","801":"snorkel","802":"snowmobile","803":"snowplow, snowplough","804":"soap dispenser","805":"soccer ball","806":"sock","807":"solar dish, solar collector, solar furnace","808":"sombrero","809":"soup bowl","810":"space bar","811":"space heater","812":"space shuttle","813":"spatula","814":"speedboat","815":"spider web, spider's web","816":"spindle","817":"sports car, sport car","818":"spotlight, spot","819":"stage","820":"steam locomotive","821":"steel arch bridge","822":"steel drum","823":"stethoscope","824":"stole","825":"stone wall","826":"stopwatch, stop watch","827":"stove","828":"strainer","829":"streetcar, tram, tramcar, trolley, trolley car","830":"stretcher","831":"studio couch, day bed","832":"stupa, tope","833":"submarine, pigboat, sub, U-boat","834":"suit, suit of clothes","835":"sundial","836":"sunglass","837":"sunglasses, dark glasses, shades","838":"sunscreen, sunblock, sun blocker","839":"suspension bridge","840":"swab, swob, mop","841":"sweatshirt","842":"swimming trunks, bathing trunks","843":"swing","844":"switch, electric switch, electrical switch","845":"syringe","846":"table lamp","847":"tank, army tank, armored combat vehicle, armoured combat vehicle","848":"tape player","849":"teapot","850":"teddy, teddy bear","851":"television, television system","852":"tennis ball","853":"thatch, thatched roof","854":"theater curtain, theatre curtain","855":"thimble","856":"thresher, thrasher, threshing machine","857":"throne","858":"tile roof","859":"toaster","860":"tobacco shop, tobacconist shop, tobacconist","861":"toilet seat","862":"torch","863":"totem pole","864":"tow truck, tow car, wrecker","865":"toyshop","866":"tractor","867":"trailer truck, tractor trailer, trucking rig, rig, articulated lorry, semi","868":"tray","869":"trench coat","870":"tricycle, trike, velocipede","871":"trimaran","872":"tripod","873":"triumphal arch","874":"trolleybus, trolley coach, trackless trolley","875":"trombone","876":"tub, vat","877":"turnstile","878":"typewriter keyboard","879":"umbrella","880":"unicycle, monocycle","881":"upright, upright piano","882":"vacuum, vacuum cleaner","883":"vase","884":"vault","885":"velvet","886":"vending machine","887":"vestment","888":"viaduct","889":"violin, fiddle","890":"volleyball","891":"waffle iron","892":"wall clock","893":"wallet, billfold, notecase, pocketbook","894":"wardrobe, closet, press","895":"warplane, military plane","896":"washbasin, handbasin, washbowl, lavabo, wash-hand basin","897":"washer, automatic washer, washing machine","898":"water bottle","899":"water jug","900":"water tower","901":"whiskey jug","902":"whistle","903":"wig","904":"window screen","905":"window shade","906":"Windsor tie","907":"wine bottle","908":"wing","909":"wok","910":"wooden spoon","911":"wool, woolen, woollen","912":"worm fence, snake fence, snake-rail fence, Virginia fence","913":"wreck","914":"yawl","915":"yurt","916":"web site, website, internet site, site","917":"comic book","918":"crossword puzzle, crossword","919":"street sign","920":"traffic light, traffic signal, stoplight","921":"book jacket, dust cover, dust jacket, dust wrapper","922":"menu","923":"plate","924":"guacamole","925":"consomme","926":"hot pot, hotpot","927":"trifle","928":"ice cream, icecream","929":"ice lolly, lolly, lollipop, popsicle","930":"French loaf","931":"bagel, beigel","932":"pretzel","933":"cheeseburger","934":"hotdog, hot dog, red hot","935":"mashed potato","936":"head cabbage","937":"broccoli","938":"cauliflower","939":"zucchini, courgette","940":"spaghetti squash","941":"acorn squash","942":"butternut squash","943":"cucumber, cuke","944":"artichoke, globe artichoke","945":"bell pepper","946":"cardoon","947":"mushroom","948":"Granny Smith","949":"strawberry","950":"orange","951":"lemon","952":"fig","953":"pineapple, ananas","954":"banana","955":"jackfruit, jak, jack","956":"custard apple","957":"pomegranate","958":"hay","959":"carbonara","960":"chocolate sauce, chocolate syrup","961":"dough","962":"meat loaf, meatloaf","963":"pizza, pizza pie","964":"potpie","965":"burrito","966":"red wine","967":"espresso","968":"cup","969":"eggnog","970":"alp","971":"bubble","972":"cliff, drop, drop-off","973":"coral reef","974":"geyser","975":"lakeside, lakeshore","976":"promontory, headland, head, foreland","977":"sandbar, sand bar","978":"seashore, coast, seacoast, sea-coast","979":"valley, vale","980":"volcano","981":"ballplayer, baseball player","982":"groom, bridegroom","983":"scuba diver","984":"rapeseed","985":"daisy","986":"yellow lady's slipper, yellow lady-slipper, Cypripedium calceolus, Cypripedium parviflorum","987":"corn","988":"acorn","989":"hip, rose hip, rosehip","990":"buckeye, horse chestnut, conker","991":"coral fungus","992":"agaric","993":"gyromitra","994":"stinkhorn, carrion fungus","995":"earthstar","996":"hen-of-the-woods, hen of the woods, Polyporus frondosus, Grifola frondosa","997":"bolete","998":"ear, spike, capitulum","999":"toilet tissue, toilet paper, bathroom tissue"}
diff --git a/model/mobilenet.json b/model/mobilenet.json
new file mode 100644
index 0000000..ef109ee
--- /dev/null
+++ b/model/mobilenet.json
@@ -0,0 +1 @@
+{"modelTopology": {"keras_version": "2.1.4", "model_config": {"class_name": "Model", "config": {"layers": [{"class_name": "InputLayer", "inbound_nodes": [], "config": {"dtype": "float32", "batch_input_shape": [null, 224, 224, 3], "name": "input_1", "sparse": false}, "name": "input_1"}, {"class_name": "Conv2D", "inbound_nodes": [[["input_1", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv1", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [2, 2], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 8, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv1"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv1", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv1_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv1_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv1_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv1_relu"}, "name": "conv1_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv1_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_1", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_1"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_1", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_1_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_1_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_1_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_1_relu"}, "name": "conv_dw_1_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_1_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_1", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 16, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_1"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_1", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_1_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_1_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_1_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_1_relu"}, "name": "conv_pw_1_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_1_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_2", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [2, 2], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_2"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_2", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_2_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_2_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_2_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_2_relu"}, "name": "conv_dw_2_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_2_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_2", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 32, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_2"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_2", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_2_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_2_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_2_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_2_relu"}, "name": "conv_pw_2_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_2_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_3", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_3"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_3", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_3_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_3_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_3_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_3_relu"}, "name": "conv_dw_3_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_3_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_3", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 32, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_3"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_3", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_3_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_3_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_3_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_3_relu"}, "name": "conv_pw_3_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_3_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_4", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [2, 2], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_4"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_4", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_4_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_4_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_4_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_4_relu"}, "name": "conv_dw_4_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_4_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_4", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 64, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_4"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_4", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_4_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_4_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_4_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_4_relu"}, "name": "conv_pw_4_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_4_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_5", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_5"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_5", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_5_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_5_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_5_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_5_relu"}, "name": "conv_dw_5_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_5_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_5", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 64, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_5"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_5", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_5_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_5_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_5_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_5_relu"}, "name": "conv_pw_5_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_5_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_6", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [2, 2], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_6"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_6", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_6_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_6_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_6_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_6_relu"}, "name": "conv_dw_6_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_6_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_6", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_6"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_6", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_6_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_6_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_6_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_6_relu"}, "name": "conv_pw_6_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_6_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_7", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_7"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_7", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_7_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_7_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_7_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_7_relu"}, "name": "conv_dw_7_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_7_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_7", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_7"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_7", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_7_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_7_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_7_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_7_relu"}, "name": "conv_pw_7_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_7_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_8", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_8"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_8", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_8_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_8_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_8_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_8_relu"}, "name": "conv_dw_8_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_8_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_8", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_8"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_8", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_8_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_8_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_8_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_8_relu"}, "name": "conv_pw_8_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_8_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_9", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_9"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_9", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_9_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_9_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_9_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_9_relu"}, "name": "conv_dw_9_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_9_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_9", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_9"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_9", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_9_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_9_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_9_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_9_relu"}, "name": "conv_pw_9_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_9_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_10", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_10"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_10", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_10_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_10_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_10_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_10_relu"}, "name": "conv_dw_10_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_10_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_10", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_10"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_10", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_10_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_10_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_10_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_10_relu"}, "name": "conv_pw_10_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_10_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_11", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_11"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_11", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_11_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_11_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_11_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_11_relu"}, "name": "conv_dw_11_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_11_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_11", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 128, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_11"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_11", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_11_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_11_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_11_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_11_relu"}, "name": "conv_pw_11_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_11_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_12", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [2, 2], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_12"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_12", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_12_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_12_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_12_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_12_relu"}, "name": "conv_dw_12_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_12_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_12", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 256, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_12"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_12", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_12_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_12_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_12_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_12_relu"}, "name": "conv_pw_12_relu"}, {"class_name": "DepthwiseConv2D", "inbound_nodes": [[["conv_pw_12_relu", 0, 0, {}]]], "config": {"padding": "same", "depth_multiplier": 1, "name": "conv_dw_13", "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "depthwise_constraint": null, "strides": [1, 1], "dilation_rate": [1, 1], "depthwise_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "depthwise_regularizer": null, "use_bias": false, "activity_regularizer": null, "kernel_size": [3, 3]}, "name": "conv_dw_13"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_dw_13", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_dw_13_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_dw_13_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_dw_13_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_dw_13_relu"}, "name": "conv_dw_13_relu"}, {"class_name": "Conv2D", "inbound_nodes": [[["conv_dw_13_relu", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_pw_13", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 256, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": false, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_pw_13"}, {"class_name": "BatchNormalization", "inbound_nodes": [[["conv_pw_13", 0, 0, {}]]], "config": {"gamma_initializer": {"class_name": "Ones", "config": {}}, "moving_mean_initializer": {"class_name": "Zeros", "config": {}}, "name": "conv_pw_13_bn", "epsilon": 0.001, "trainable": true, "center": true, "moving_variance_initializer": {"class_name": "Ones", "config": {}}, "beta_initializer": {"class_name": "Zeros", "config": {}}, "scale": true, "gamma_regularizer": null, "gamma_constraint": null, "beta_constraint": null, "beta_regularizer": null, "momentum": 0.99, "axis": -1}, "name": "conv_pw_13_bn"}, {"class_name": "Activation", "inbound_nodes": [[["conv_pw_13_bn", 0, 0, {}]]], "config": {"activation": "relu6", "trainable": true, "name": "conv_pw_13_relu"}, "name": "conv_pw_13_relu"}, {"class_name": "GlobalAveragePooling2D", "inbound_nodes": [[["conv_pw_13_relu", 0, 0, {}]]], "config": {"trainable": true, "name": "global_average_pooling2d_1", "data_format": "channels_last"}, "name": "global_average_pooling2d_1"}, {"class_name": "Reshape", "inbound_nodes": [[["global_average_pooling2d_1", 0, 0, {}]]], "config": {"target_shape": [1, 1, 256], "trainable": true, "name": "reshape_1"}, "name": "reshape_1"}, {"class_name": "Dropout", "inbound_nodes": [[["reshape_1", 0, 0, {}]]], "config": {"rate": 0.001, "noise_shape": null, "trainable": true, "seed": null, "name": "dropout"}, "name": "dropout"}, {"class_name": "Conv2D", "inbound_nodes": [[["dropout", 0, 0, {}]]], "config": {"kernel_initializer": {"class_name": "VarianceScaling", "config": {"distribution": "uniform", "scale": 1.0, "seed": null, "mode": "fan_avg"}}, "name": "conv_preds", "kernel_constraint": null, "bias_regularizer": null, "bias_constraint": null, "activation": "linear", "trainable": true, "data_format": "channels_last", "padding": "same", "strides": [1, 1], "dilation_rate": [1, 1], "kernel_regularizer": null, "filters": 1000, "bias_initializer": {"class_name": "Zeros", "config": {}}, "use_bias": true, "activity_regularizer": null, "kernel_size": [1, 1]}, "name": "conv_preds"}, {"class_name": "Activation", "inbound_nodes": [[["conv_preds", 0, 0, {}]]], "config": {"activation": "softmax", "trainable": true, "name": "act_softmax"}, "name": "act_softmax"}, {"class_name": "Reshape", "inbound_nodes": [[["act_softmax", 0, 0, {}]]], "config": {"target_shape": [1000], "trainable": true, "name": "reshape_2"}, "name": "reshape_2"}], "input_layers": [["input_1", 0, 0]], "name": "mobilenet_0.25_224", "output_layers": [["reshape_2", 0, 0]]}}, "backend": "tensorflow"}, "weightsManifest": [{"paths": ["group1-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 3, 8], "name": "conv1/kernel"}]}, {"paths": ["group2-shard1of1"], "weights": [{"dtype": "float32", "shape": [8], "name": "conv1_bn/gamma"}, {"dtype": "float32", "shape": [8], "name": "conv1_bn/beta"}, {"dtype": "float32", "shape": [8], "name": "conv1_bn/moving_mean"}, {"dtype": "float32", "shape": [8], "name": "conv1_bn/moving_variance"}]}, {"paths": ["group3-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 8, 1], "name": "conv_dw_1/depthwise_kernel"}]}, {"paths": ["group4-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_10/depthwise_kernel"}]}, {"paths": ["group5-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_10_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_10_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_10_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_10_bn/moving_variance"}]}, {"paths": ["group6-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_11/depthwise_kernel"}]}, {"paths": ["group7-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_11_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_11_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_11_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_11_bn/moving_variance"}]}, {"paths": ["group8-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_12/depthwise_kernel"}]}, {"paths": ["group9-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_12_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_12_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_12_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_12_bn/moving_variance"}]}, {"paths": ["group10-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 256, 1], "name": "conv_dw_13/depthwise_kernel"}]}, {"paths": ["group11-shard1of1"], "weights": [{"dtype": "float32", "shape": [256], "name": "conv_dw_13_bn/gamma"}, {"dtype": "float32", "shape": [256], "name": "conv_dw_13_bn/beta"}, {"dtype": "float32", "shape": [256], "name": "conv_dw_13_bn/moving_mean"}, {"dtype": "float32", "shape": [256], "name": "conv_dw_13_bn/moving_variance"}]}, {"paths": ["group12-shard1of1"], "weights": [{"dtype": "float32", "shape": [8], "name": "conv_dw_1_bn/gamma"}, {"dtype": "float32", "shape": [8], "name": "conv_dw_1_bn/beta"}, {"dtype": "float32", "shape": [8], "name": "conv_dw_1_bn/moving_mean"}, {"dtype": "float32", "shape": [8], "name": "conv_dw_1_bn/moving_variance"}]}, {"paths": ["group13-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 16, 1], "name": "conv_dw_2/depthwise_kernel"}]}, {"paths": ["group14-shard1of1"], "weights": [{"dtype": "float32", "shape": [16], "name": "conv_dw_2_bn/gamma"}, {"dtype": "float32", "shape": [16], "name": "conv_dw_2_bn/beta"}, {"dtype": "float32", "shape": [16], "name": "conv_dw_2_bn/moving_mean"}, {"dtype": "float32", "shape": [16], "name": "conv_dw_2_bn/moving_variance"}]}, {"paths": ["group15-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 32, 1], "name": "conv_dw_3/depthwise_kernel"}]}, {"paths": ["group16-shard1of1"], "weights": [{"dtype": "float32", "shape": [32], "name": "conv_dw_3_bn/gamma"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_3_bn/beta"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_3_bn/moving_mean"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_3_bn/moving_variance"}]}, {"paths": ["group17-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 32, 1], "name": "conv_dw_4/depthwise_kernel"}]}, {"paths": ["group18-shard1of1"], "weights": [{"dtype": "float32", "shape": [32], "name": "conv_dw_4_bn/gamma"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_4_bn/beta"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_4_bn/moving_mean"}, {"dtype": "float32", "shape": [32], "name": "conv_dw_4_bn/moving_variance"}]}, {"paths": ["group19-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 64, 1], "name": "conv_dw_5/depthwise_kernel"}]}, {"paths": ["group20-shard1of1"], "weights": [{"dtype": "float32", "shape": [64], "name": "conv_dw_5_bn/gamma"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_5_bn/beta"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_5_bn/moving_mean"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_5_bn/moving_variance"}]}, {"paths": ["group21-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 64, 1], "name": "conv_dw_6/depthwise_kernel"}]}, {"paths": ["group22-shard1of1"], "weights": [{"dtype": "float32", "shape": [64], "name": "conv_dw_6_bn/gamma"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_6_bn/beta"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_6_bn/moving_mean"}, {"dtype": "float32", "shape": [64], "name": "conv_dw_6_bn/moving_variance"}]}, {"paths": ["group23-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_7/depthwise_kernel"}]}, {"paths": ["group24-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_7_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_7_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_7_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_7_bn/moving_variance"}]}, {"paths": ["group25-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_8/depthwise_kernel"}]}, {"paths": ["group26-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_8_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_8_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_8_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_8_bn/moving_variance"}]}, {"paths": ["group27-shard1of1"], "weights": [{"dtype": "float32", "shape": [3, 3, 128, 1], "name": "conv_dw_9/depthwise_kernel"}]}, {"paths": ["group28-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_dw_9_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_9_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_9_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_dw_9_bn/moving_variance"}]}, {"paths": ["group29-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 256, 1000], "name": "conv_preds/kernel"}, {"dtype": "float32", "shape": [1000], "name": "conv_preds/bias"}]}, {"paths": ["group30-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 8, 16], "name": "conv_pw_1/kernel"}]}, {"paths": ["group31-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 128], "name": "conv_pw_10/kernel"}]}, {"paths": ["group32-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_10_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_10_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_10_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_10_bn/moving_variance"}]}, {"paths": ["group33-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 128], "name": "conv_pw_11/kernel"}]}, {"paths": ["group34-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_11_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_11_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_11_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_11_bn/moving_variance"}]}, {"paths": ["group35-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 256], "name": "conv_pw_12/kernel"}]}, {"paths": ["group36-shard1of1"], "weights": [{"dtype": "float32", "shape": [256], "name": "conv_pw_12_bn/gamma"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_12_bn/beta"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_12_bn/moving_mean"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_12_bn/moving_variance"}]}, {"paths": ["group37-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 256, 256], "name": "conv_pw_13/kernel"}]}, {"paths": ["group38-shard1of1"], "weights": [{"dtype": "float32", "shape": [256], "name": "conv_pw_13_bn/gamma"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_13_bn/beta"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_13_bn/moving_mean"}, {"dtype": "float32", "shape": [256], "name": "conv_pw_13_bn/moving_variance"}]}, {"paths": ["group39-shard1of1"], "weights": [{"dtype": "float32", "shape": [16], "name": "conv_pw_1_bn/gamma"}, {"dtype": "float32", "shape": [16], "name": "conv_pw_1_bn/beta"}, {"dtype": "float32", "shape": [16], "name": "conv_pw_1_bn/moving_mean"}, {"dtype": "float32", "shape": [16], "name": "conv_pw_1_bn/moving_variance"}]}, {"paths": ["group40-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 16, 32], "name": "conv_pw_2/kernel"}]}, {"paths": ["group41-shard1of1"], "weights": [{"dtype": "float32", "shape": [32], "name": "conv_pw_2_bn/gamma"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_2_bn/beta"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_2_bn/moving_mean"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_2_bn/moving_variance"}]}, {"paths": ["group42-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 32, 32], "name": "conv_pw_3/kernel"}]}, {"paths": ["group43-shard1of1"], "weights": [{"dtype": "float32", "shape": [32], "name": "conv_pw_3_bn/gamma"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_3_bn/beta"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_3_bn/moving_mean"}, {"dtype": "float32", "shape": [32], "name": "conv_pw_3_bn/moving_variance"}]}, {"paths": ["group44-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 32, 64], "name": "conv_pw_4/kernel"}]}, {"paths": ["group45-shard1of1"], "weights": [{"dtype": "float32", "shape": [64], "name": "conv_pw_4_bn/gamma"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_4_bn/beta"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_4_bn/moving_mean"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_4_bn/moving_variance"}]}, {"paths": ["group46-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 64, 64], "name": "conv_pw_5/kernel"}]}, {"paths": ["group47-shard1of1"], "weights": [{"dtype": "float32", "shape": [64], "name": "conv_pw_5_bn/gamma"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_5_bn/beta"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_5_bn/moving_mean"}, {"dtype": "float32", "shape": [64], "name": "conv_pw_5_bn/moving_variance"}]}, {"paths": ["group48-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 64, 128], "name": "conv_pw_6/kernel"}]}, {"paths": ["group49-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_6_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_6_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_6_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_6_bn/moving_variance"}]}, {"paths": ["group50-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 128], "name": "conv_pw_7/kernel"}]}, {"paths": ["group51-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_7_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_7_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_7_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_7_bn/moving_variance"}]}, {"paths": ["group52-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 128], "name": "conv_pw_8/kernel"}]}, {"paths": ["group53-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_8_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_8_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_8_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_8_bn/moving_variance"}]}, {"paths": ["group54-shard1of1"], "weights": [{"dtype": "float32", "shape": [1, 1, 128, 128], "name": "conv_pw_9/kernel"}]}, {"paths": ["group55-shard1of1"], "weights": [{"dtype": "float32", "shape": [128], "name": "conv_pw_9_bn/gamma"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_9_bn/beta"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_9_bn/moving_mean"}, {"dtype": "float32", "shape": [128], "name": "conv_pw_9_bn/moving_variance"}]}]}
\ No newline at end of file
diff --git a/notification.ogg b/notification.ogg
new file mode 100644
index 0000000..2f513ef
Binary files /dev/null and b/notification.ogg differ
diff --git a/roboto-black.ttf b/roboto-black.ttf
new file mode 100644
index 0000000..689fe5c
Binary files /dev/null and b/roboto-black.ttf differ
diff --git a/tensorflow.js b/tensorflow.js
new file mode 100644
index 0000000..a2a4c67
--- /dev/null
+++ b/tensorflow.js
@@ -0,0 +1 @@
+!function(e){if("object"==typeof exports&&"undefined"!=typeof module)module.exports=e();else if("function"==typeof define&&define.amd)define([],e);else{("undefined"!=typeof window?window:"undefined"!=typeof global?global:"undefined"!=typeof self?self:this).tf=e()}}(function(){return function(){function e(t,n,r){function i(o,s){if(!n[o]){if(!t[o]){var u="function"==typeof require&&require;if(!s&&u)return u(o,!0);if(a)return a(o,!0);var l=new Error("Cannot find module '"+o+"'");throw l.code="MODULE_NOT_FOUND",l}var c=n[o]={exports:{}};t[o][0].call(c.exports,function(e){var n=t[o][1][e];return i(n||e)},c,c.exports,e,t,n,r)}return n[o].exports}for(var a="function"==typeof require&&require,o=0;o<r.length;o++)i(r[o]);return i}return e}()({1:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("./doc"),a=function(){function e(){}return e.nextFrame=function(){return new Promise(function(e){return requestAnimationFrame(function(){return e()})})},r([i.doc({heading:"Performance",subheading:"Timing"})],e,"nextFrame",null),e}();n.BrowserUtil=a},{"./doc":3}],2:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0}),n.isMobile=function(){var e=navigator.userAgent||navigator.vendor||window.opera;return/(android|bb\d+|meego).+mobile|avantgo|bada\/|blackberry|blazer|compal|elaine|fennec|hiptop|iemobile|ip(hone|od)|iris|kindle|lge |maemo|midp|mmp|mobile.+firefox|netfront|opera m(ob|in)i|palm( os)?|phone|p(ixi|re)\/|plucker|pocket|psp|series(4|6)0|symbian|treo|up\.(browser|link)|vodafone|wap|windows ce|xda|xiino/i.test(e)||/1207|6310|6590|3gso|4thp|50[1-6]i|770s|802s|a wa|abac|ac(er|oo|s\-)|ai(ko|rn)|al(av|ca|co)|amoi|an(ex|ny|yw)|aptu|ar(ch|go)|as(te|us)|attw|au(di|\-m|r |s )|avan|be(ck|ll|nq)|bi(lb|rd)|bl(ac|az)|br(e|v)w|bumb|bw\-(n|u)|c55\/|capi|ccwa|cdm\-|cell|chtm|cldc|cmd\-|co(mp|nd)|craw|da(it|ll|ng)|dbte|dc\-s|devi|dica|dmob|do(c|p)o|ds(12|\-d)|el(49|ai)|em(l2|ul)|er(ic|k0)|esl8|ez([4-7]0|os|wa|ze)|fetc|fly(\-|_)|g1 u|g560|gene|gf\-5|g\-mo|go(\.w|od)|gr(ad|un)|haie|hcit|hd\-(m|p|t)|hei\-|hi(pt|ta)|hp( i|ip)|hs\-c|ht(c(\-| |_|a|g|p|s|t)|tp)|hu(aw|tc)|i\-(20|go|ma)|i230|iac( |\-|\/)|ibro|idea|ig01|ikom|im1k|inno|ipaq|iris|ja(t|v)a|jbro|jemu|jigs|kddi|keji|kgt( |\/)|klon|kpt |kwc\-|kyo(c|k)|le(no|xi)|lg( g|\/(k|l|u)|50|54|\-[a-w])|libw|lynx|m1\-w|m3ga|m50\/|ma(te|ui|xo)|mc(01|21|ca)|m\-cr|me(rc|ri)|mi(o8|oa|ts)|mmef|mo(01|02|bi|de|do|t(\-| |o|v)|zz)|mt(50|p1|v )|mwbp|mywa|n10[0-2]|n20[2-3]|n30(0|2)|n50(0|2|5)|n7(0(0|1)|10)|ne((c|m)\-|on|tf|wf|wg|wt)|nok(6|i)|nzph|o2im|op(ti|wv)|oran|owg1|p800|pan(a|d|t)|pdxg|pg(13|\-([1-8]|c))|phil|pire|pl(ay|uc)|pn\-2|po(ck|rt|se)|prox|psio|pt\-g|qa\-a|qc(07|12|21|32|60|\-[2-7]|i\-)|qtek|r380|r600|raks|rim9|ro(ve|zo)|s55\/|sa(ge|ma|mm|ms|ny|va)|sc(01|h\-|oo|p\-)|sdk\/|se(c(\-|0|1)|47|mc|nd|ri)|sgh\-|shar|sie(\-|m)|sk\-0|sl(45|id)|sm(al|ar|b3|it|t5)|so(ft|ny)|sp(01|h\-|v\-|v )|sy(01|mb)|t2(18|50)|t6(00|10|18)|ta(gt|lk)|tcl\-|tdg\-|tel(i|m)|tim\-|t\-mo|to(pl|sh)|ts(70|m\-|m3|m5)|tx\-9|up(\.b|g1|si)|utst|v400|v750|veri|vi(rg|te)|vk(40|5[0-3]|\-v)|vm40|voda|vulc|vx(52|53|60|61|70|80|81|83|85|98)|w3c(\-| )|webc|whit|wi(g |nc|nw)|wmlb|wonu|x700|yas\-|your|zeto|zte\-/i.test(e.substr(0,4))}},{}],3:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0}),n.doc=function(e){return function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t]}}},{}],4:[function(e,t,n){"use strict";var r=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},i=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var a=e("./environment"),o=e("./globals"),s=e("./ops/ops"),u=e("./profiler"),l=e("./tape"),c=e("./tensor"),p=e("./util"),h=function(){function e(e,t){this.backend=e,this.safeMode=t,this.registeredVariables={},this.refCounter=new WeakMap,this.nextTapeNodeId=0,this.numBytes=0,this.numTensors=0,this.numDataBuffers=0,this.gradientScopeCount=0,this.customGradientDepth=0,this.activeScope={keep:[],track:[]},this.scopeStack=[this.activeScope],this.profiler=new u.Profiler(e)}return e.prototype.runKernel=function(e,t,n){var r,i=this,o=[],s=function(e){return o.push(e),e},u=this.activeScope.name;if(this.customGradientDepth++,r=a.ENV.get("DEBUG")?this.profiler.profileKernel(u,function(){return e(i.backend,s)}):e(this.backend,s),this.customGradientDepth--,this.shouldRecord()){var l={id:this.nextTapeNodeId++,name:u,inputs:t,output:r};null!=n&&(l.gradient=function(e){return n(e,o)}),this.activeTape.push(l)}return r},e.prototype.registerTensor=function(e){var t=this.refCounter.has(e.dataId)?this.refCounter.get(e.dataId):0;this.numTensors++,0===t&&(this.numDataBuffers++,this.numBytes+=p.sizeFromShape(e.shape)*p.bytesPerElement(e.dtype),this.backend.register(e.dataId,e.shape,e.dtype)),this.refCounter.set(e.dataId,t+1),e instanceof c.Variable||this.track(e)},e.prototype.registerVariable=function(e){if(null!=this.registeredVariables[e.name])throw new Error("Variable with name "+e.name+" was already registered");this.registeredVariables[e.name]=e},e.prototype.disposeTensor=function(e){if(this.refCounter.has(e.dataId)){this.numTensors--;var t=this.refCounter.get(e.dataId);t<=1?(this.refCounter.delete(e.dataId),this.backend.disposeData(e.dataId),this.numDataBuffers--,this.numBytes-=p.sizeFromShape(e.shape)*p.bytesPerElement(e.dtype)):this.refCounter.set(e.dataId,t-1)}},e.prototype.memory=function(){var e=this.backend.memory();return e.numTensors=this.numTensors,e.numDataBuffers=this.numDataBuffers,e.numBytes=this.numBytes,e},e.prototype.shouldRecord=function(){return null!=this.activeTape&&0===this.customGradientDepth},e.prototype.addTapeNode=function(e,t,n){var r={};e.forEach(function(e,t){r[t]=e});var i={id:this.nextTapeNodeId++,name:this.activeScope.name,inputs:r,output:t,gradient:function(e){var t={};return n(e).forEach(function(e,n){t[n]=function(){return e}}),t}};this.activeTape.push(i)},e.prototype.keep=function(e){if(1===this.scopeStack.length&&a.ENV.engine.safeMode)throw new Error("Safe mode is ON. Enclose all tensor operations inside tf.tidy(): tf.tidy(() => {...}) to avoid memory leaks.");return this.activeScope.keep.push(e),e},e.prototype.startScope=function(e,t){void 0===t&&(t=!1),t&&0===this.gradientScopeCount&&(this.activeTape=[]),t&&this.gradientScopeCount++;var n={keep:[],track:[]};e&&(n.name=e),this.scopeStack.push(n),this.activeScope=n},e.prototype.endScope=function(e,t){var n=this;void 0===t&&(t=!1),t&&0===--this.gradientScopeCount&&(this.activeTape=null);var r=this.activeScope.keep,i=p.extractTensorsFromContainer(e);r=r.concat(i);for(var a=0;a<this.activeScope.track.length;a++){var o=this.activeScope.track[a];p.isTensorInList(o,r)||(null!=this.activeTape?i.push(o):o.dispose())}this.scopeStack.pop(),this.activeScope=0===this.scopeStack.length?{keep:[],track:[]}:this.scopeStack[this.scopeStack.length-1],i.forEach(function(e){p.isTensorInList(e,n.activeScope.keep)||n.track(e)})},e.prototype.dispose=function(){},e.prototype.gradients=function(e,t,n,r){var i=this;return void 0===r&&(r=!1),p.assert(t.length>0,"gradients() received an empty list of xs."),o.tidy("gradients",function(){var a=e();p.assert(a instanceof c.Tensor,"The result y returned by f() must be a tensor.");var o=l.getFilteredNodesXToY(i.activeTape,t,a);if(!r&&0===o.length&&t.length>0)throw new Error("Cannot compute gradient of y=f(x) with respect to x. Make sure that the f you passed encloses all operations that lead from x to y.");var u={};return u[a.id]=null==n?s.ones(a.shape):n,l.backpropagateGradients(u,o),{value:a,grads:t.map(function(e){return u[e.id]})}},!0)},e.prototype.customGrad=function(e){var t=this;return p.assert(p.isFunction(e),"The f passed in customGrad(f) must be a function."),function(){for(var n=[],r=0;r<arguments.length;r++)n[r]=arguments[r];p.assert(n.every(function(e){return e instanceof c.Tensor}),"The args passed in customGrad(f)(x1, x2,...) must all be tensors"),t.customGradientDepth++;var i,a=o.tidy(e.name,function(){var t=e.apply(void 0,n),r=t.value,a=t.gradFunc;return p.assert(r instanceof c.Tensor,"The function f passed in customGrad(f) must return an object where `obj.value` is a tensor"),p.assert(p.isFunction(a),"The function f passed in customGrad(f) must return an object where `obj.gradFunc` is a function."),i=a,r},!0);if(t.customGradientDepth--,t.shouldRecord()){t.addTapeNode(n,a,function(e){var t=i(e),r=Array.isArray(t)?t:[t];return p.assert(r.length===n.length,"The function f passed in customGrad(f) must return an object where `obj.gradFunc` is a function that returns the same number of tensors as inputs passed to f(...)."),p.assert(r.every(function(e){return e instanceof c.Tensor}),"The function f passed in customGrad(f) must return an object where `obj.gradFunc` is a function that returns a list of only tensors."),r})}return a}},e.prototype.write=function(e,t){this.backend.write(e,t)},e.prototype.readSync=function(e){return this.backend.readSync(e)},e.prototype.read=function(e){return this.backend.read(e)},e.prototype.fromPixels=function(e,t){return this.backend.fromPixels(e,t)},e.prototype.time=function(e){return r(this,void 0,void 0,function(){var t,n;return i(this,function(r){switch(r.label){case 0:return t=performance.now(),[4,this.backend.time(e)];case 1:return n=r.sent(),n.wallMs=performance.now()-t,[2,n]}})})},e.prototype.track=function(e){if(1===this.scopeStack.length&&this.safeMode)throw new Error("Safe mode is ON. Enclose all tensor operations inside tf.tidy(): tf.tidy(() => {op();...}); to avoid memory leaks.");return this.activeScope.track.push(e),e},e}();n.Engine=h},{"./environment":5,"./globals":6,"./ops/ops":66,"./profiler":87,"./tape":88,"./tensor":89,"./util":95}],5:[function(e,t,n){(function(t){"use strict";function r(e,t){return null!=e.getExtension(t)}function i(e){if(0===e)throw new Error("Cannot get WebGL rendering context, WebGL is disabled.");var t=document.createElement("canvas");return 1===e?t.getContext("webgl")||t.getContext("experimental-webgl"):t.getContext("webgl2")}function a(e){if(null!=e){var t=e.getExtension("WEBGL_lose_context");if(null==t)throw new Error("Extension WEBGL_lose_context not supported on this browser.");t.loseContext()}}function o(e){var t=i(e);return null!=t&&(a(t),!0)}function s(e){if(0===e)return 0;var t,n=i(e);return t=r(n,"EXT_disjoint_timer_query_webgl2")&&2===e?2:r(n,"EXT_disjoint_timer_query")?1:0,null!=n&&a(n),t}function u(e){if(0===e)return!1;var t=i(e);if(1===e){if(!r(t,"OES_texture_float"))return!1}else if(!r(t,"EXT_color_buffer_float"))return!1;var n=t.createFramebuffer(),o=t.createTexture();t.bindTexture(t.TEXTURE_2D,o);var s=2===e?t.RGBA32F:t.RGBA;t.texImage2D(t.TEXTURE_2D,0,s,1,1,0,t.RGBA,t.FLOAT,null),t.bindFramebuffer(t.FRAMEBUFFER,n),t.framebufferTexture2D(t.FRAMEBUFFER,t.COLOR_ATTACHMENT0,t.TEXTURE_2D,o,0);var u=t.checkFramebufferStatus(t.FRAMEBUFFER)===t.FRAMEBUFFER_COMPLETE;t.readPixels(0,0,1,1,t.RGBA,t.FLOAT,new Float32Array(4));var l=t.getError()===t.NO_ERROR;return a(t),u&&l}function l(e){if(e>0)return!1;if(2!==e)return!1;var t=i(e),n=r(t,"WEBGL_get_buffer_sub_data_async");return a(t),n}function c(){var e={};if("undefined"==typeof window)return e;var t=v.getQueryParams(window.location.search);if(b in t){var r={};t[b].split(",").forEach(function(e){var t=e.split(":"),n=t[0],i=t[1];r[n]=i}),n.URL_PROPERTIES.forEach(function(t){t.name in r&&(console.log("Setting feature override from URL "+t.name+": "+r[t.name]),t.type===d.NUMBER?e[t.name]=+r[t.name]:t.type===d.BOOLEAN?e[t.name]="true"===r[t.name]:t.type===d.STRING?e[t.name]=r[t.name]:console.warn("Unknown URL param: "+t.name+"."))})}return e}function p(){var e;if("undefined"!=typeof window)e=window;else{if(void 0===t)throw new Error("Could not find a global object");e=t}return e}var h=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var d,f=e("./device_util"),g=e("./doc"),m=e("./engine"),v=e("./util");!function(e){e[e.NUMBER=0]="NUMBER",e[e.BOOLEAN=1]="BOOLEAN",e[e.STRING=2]="STRING"}(d=n.Type||(n.Type={})),n.URL_PROPERTIES=[{name:"DEBUG",type:d.BOOLEAN},{name:"WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION",type:d.NUMBER},{name:"WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_RELIABLE",type:d.BOOLEAN},{name:"WEBGL_VERSION",type:d.NUMBER},{name:"WEBGL_FLOAT_TEXTURE_ENABLED",type:d.BOOLEAN},{name:"WEBGL_GET_BUFFER_SUB_DATA_ASYNC_EXTENSION_ENABLED",type:d.BOOLEAN},{name:"BACKEND",type:d.STRING}];var y=function(){function e(e){this.features={},this.registry={},null!=e&&(this.features=e),this.get("DEBUG")&&console.warn("Debugging mode is ON. The output of every math call will be downloaded to CPU and checked for NaNs. This significantly impacts performance.")}return e.setBackend=function(e,t){if(void 0===t&&(t=!1),!(e in n.ENV.registry))throw new Error("Backend type '"+e+"' not found in registry");n.ENV.initBackend(e,t)},e.getBackend=function(){return n.ENV.initDefaultBackend(),n.ENV.currentBackend},e.memory=function(){return n.ENV.engine.memory()},e.prototype.get=function(e){return e in this.features?this.features[e]:(this.features[e]=this.evaluateFeature(e),this.features[e])},e.prototype.set=function(e,t){this.features[e]=t},e.prototype.getBestBackendType=function(){var e=this;if(0===Object.keys(this.registry).length)throw new Error("No backend found in registry.");return Object.keys(this.registry).map(function(t){return{name:t,entry:e.registry[t]}}).sort(function(e,t){return t.entry.priority-e.entry.priority})[0].name},e.prototype.evaluateFeature=function(e){if("DEBUG"===e)return!1;if("BACKEND"===e)return this.getBestBackendType();if("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION"===e){var t=this.get("WEBGL_VERSION");return 0===t?0:s(t)}if("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_RELIABLE"===e)return this.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")>0&&!f.isMobile();if("WEBGL_VERSION"===e)return o(2)?2:o(1)?1:0;if("WEBGL_FLOAT_TEXTURE_ENABLED"===e)return u(this.get("WEBGL_VERSION"));if("WEBGL_GET_BUFFER_SUB_DATA_ASYNC_EXTENSION_ENABLED"===e)return l(this.get("WEBGL_VERSION"));throw new Error("Unknown feature "+e+".")},e.prototype.setFeatures=function(e){this.features=e},e.prototype.reset=function(){this.features=c(),null!=this.globalEngine&&(this.globalEngine.dispose(),this.globalEngine=null)},e.prototype.initBackend=function(e,t){void 0===t&&(t=!1),this.currentBackend=e,null!=this.globalEngine&&this.globalEngine.dispose();var r=n.ENV.findBackend(e);this.globalEngine=new m.Engine(r,t)},e.prototype.findBackend=function(e){return e in this.registry?this.registry[e].backend:null},e.prototype.registerBackend=function(e,t,n){void 0===n&&(n=1),e in this.registry&&console.warn(e+" backend was already registered");try{var r=t();return this.registry[e]={backend:r,priority:n},!0}catch(e){return console.warn(e.message),!1}},e.prototype.removeBackend=function(e){if(!(e in this.registry))throw new Error(e+" backend not found in registry");this.registry[e].backend.dispose(),delete this.registry[e]},Object.defineProperty(e.prototype,"engine",{get:function(){return this.initDefaultBackend(),this.globalEngine},enumerable:!0,configurable:!0}),e.prototype.initDefaultBackend=function(){null==this.globalEngine&&this.initBackend(n.ENV.get("BACKEND"),!1)},h([g.doc({heading:"Environment"})],e,"setBackend",null),h([g.doc({heading:"Environment"})],e,"getBackend",null),h([g.doc({heading:"Performance",subheading:"Memory"})],e,"memory",null),e}();n.Environment=y;var b="tfjsflags";n.ENV=function(){var e=p();return e.ENV=e.ENV||new y(c()),e.ENV}()}).call(this,"undefined"!=typeof global?global:"undefined"!=typeof self?self:"undefined"!=typeof window?window:{})},{"./device_util":2,"./doc":3,"./engine":4,"./util":95}],6:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./gradients"),i=e("./tracking");n.tidy=i.Tracking.tidy,n.keep=i.Tracking.keep,n.dispose=i.Tracking.dispose,n.time=i.Tracking.time,n.grad=r.Gradients.grad,n.valueAndGrad=r.Gradients.valueAndGrad,n.grads=r.Gradients.grads,n.valueAndGrads=r.Gradients.valueAndGrads,n.variableGrads=r.Gradients.variableGrads,n.customGrad=r.Gradients.customGrad},{"./gradients":7,"./tracking":92}],7:[function(e,t,n){"use strict";function r(e){if(e.filter(function(e){return null==e}).length>0)throw new Error("Cannot compute gradient of y=f(x) with respect to x. Make sure that\n    the f you passed encloses all operations that lead from x to y.")}var i=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var a=e("./doc"),o=e("./environment"),s=e("./globals"),u=e("./tensor"),l=e("./util"),c=function(){function e(){}return e.gradScope=function(e,t){return s.tidy(e,t,!0)},e.grad=function(e){return l.assert(l.isFunction(e),"The f passed in grad(f) must be a function"),function(t,n){l.assert(t instanceof u.Tensor,"The x passed in grad(f)(x) must be a tensor"),l.assert(null==n||n instanceof u.Tensor,"The dy passed in grad(f)(x, dy) must be a tensor");var i=o.ENV.engine.gradients(function(){return e(t)},[t],n),a=i.value,s=i.grads;return null!=n&&l.assertShapesMatch(a.shape,n.shape,"The shape of dy passed in grad(f)(x, dy) must match the shape returned by f(x)"),a.dispose(),r(s),s[0]}},e.grads=function(e){return l.assert(l.isFunction(e),"The f passed in grads(f) must be a function"),function(t,n){l.assert(Array.isArray(t)&&t.every(function(e){return e instanceof u.Tensor}),"The args passed in grads(f)(args) must be an array of tensors"),l.assert(null==n||n instanceof u.Tensor,"The dy passed in grads(f)(args, dy) must be a tensor");var i=o.ENV.engine.gradients(function(){return e.apply(void 0,t)},t,n),a=i.value,s=i.grads;return null!=n&&l.assertShapesMatch(a.shape,n.shape,"The shape of dy passed in grads(f)([x1,...], dy) must match the shape returned by f([x1,...])"),a.dispose(),r(s),s}},e.valueAndGrad=function(e){return l.assert(l.isFunction(e),"The f passed in valueAndGrad(f) must be a function"),function(t,n){l.assert(t instanceof u.Tensor,"The x passed in valueAndGrad(f)(x) must be a tensor"),l.assert(null==n||n instanceof u.Tensor,"The dy passed in valueAndGrad(f)(x, dy) must be a tensor");var i=o.ENV.engine.gradients(function(){return e(t)},[t],n),a=i.grads,s=i.value;return r(a),{grad:a[0],value:s}}},e.valueAndGrads=function(e){return l.assert(l.isFunction(e),"The f passed in valueAndGrads(f) must be a function"),function(t,n){l.assert(Array.isArray(t)&&t.every(function(e){return e instanceof u.Tensor}),"The args passed in valueAndGrads(f)(args) must be array of tensors"),l.assert(null==n||n instanceof u.Tensor,"The dy passed in valueAndGrads(f)(args, dy) must be a tensor");var i=o.ENV.engine.gradients(function(){return e.apply(void 0,t)},t,n);return null!=n&&l.assertShapesMatch(i.value.shape,n.shape,"The shape of dy passed in valueAndGrads(f)([x1,...], dy) must match the shape returned by f([x1,...])"),r(i.grads),i}},e.variableGrads=function(e,t){if(l.assert(l.isFunction(e),"The f passed in variableGrads(f) must be a function"),l.assert(null==t||Array.isArray(t)&&t.every(function(e){return e instanceof u.Variable}),"The varList passed in variableGrads(f, varList) must be an array of variables"),null==t){t=[];for(var n in o.ENV.engine.registeredVariables)t.push(o.ENV.engine.registeredVariables[n])}var r=t.length;t=t.filter(function(e){return e.trainable}),l.assert(t.length>0,"variableGrads() expects at least one of the input variables to be trainable, but none of the "+r+" variables is trainable.");var i=o.ENV.engine.gradients(e,t,null,!0),a=i.value,s=i.grads;l.assert(s.some(function(e){return null!=e}),"Cannot find a connection between any variable and the result of the loss function y=f(x). Please make sure the operations that use variables are inside the function f passed to minimize()."),l.assert(0===a.rank,"The f passed in variableGrads(f) must return a scalar, but it returned a rank-"+a.rank+" tensor");var c={};return t.forEach(function(e,t){null!=s[t]&&(c[e.name]=s[t])}),{value:a,grads:c}},e.customGrad=function(e){return o.ENV.engine.customGrad(e)},i([a.doc({heading:"Training",subheading:"Gradients"})],e,"grad",null),i([a.doc({heading:"Training",subheading:"Gradients"})],e,"grads",null),i([a.doc({heading:"Training",subheading:"Gradients"})],e,"valueAndGrad",null),i([a.doc({heading:"Training",subheading:"Gradients"})],e,"valueAndGrads",null),i([a.doc({heading:"Training",subheading:"Gradients"})],e,"variableGrads",null),i([a.doc({heading:"Training",subheading:"Gradients"})],e,"customGrad",null),e}();n.Gradients=c},{"./doc":3,"./environment":5,"./globals":6,"./tensor":89,"./util":95}],8:[function(e,t,n){"use strict";function r(e){for(var t in e)n.hasOwnProperty(t)||(n[t]=e[t])}Object.defineProperty(n,"__esModule",{value:!0}),e("./kernels/backend_webgl"),e("./kernels/backend_cpu");var i=e("./browser_util"),a=e("./environment");n.environment=a;var o=e("./environment"),s=e("./kernels/webgl/gpgpu_util"),u=e("./kernels/webgl/webgl_util"),l=e("./test_util");n.test_util=l;var c=e("./util");n.util=c;var p=e("./version");n.version_core=p.version;var h=e("./optimizers/adadelta_optimizer");n.AdadeltaOptimizer=h.AdadeltaOptimizer;var d=e("./optimizers/adagrad_optimizer");n.AdagradOptimizer=d.AdagradOptimizer;var f=e("./optimizers/adam_optimizer");n.AdamOptimizer=f.AdamOptimizer;var g=e("./optimizers/adamax_optimizer");n.AdamaxOptimizer=g.AdamaxOptimizer;var m=e("./optimizers/momentum_optimizer");n.MomentumOptimizer=m.MomentumOptimizer;var v=e("./optimizers/optimizer");n.Optimizer=v.Optimizer;var y=e("./optimizers/rmsprop_optimizer");n.RMSPropOptimizer=y.RMSPropOptimizer;var b=e("./optimizers/sgd_optimizer");n.SGDOptimizer=b.SGDOptimizer;var w=e("./tensor");n.Tensor=w.Tensor,n.TensorBuffer=w.TensorBuffer,n.variable=w.variable,n.Variable=w.Variable;var x=e("./types");n.Rank=x.Rank;var A=e("./weights_loader");n.loadWeights=A.loadWeights,r(e("./ops/ops"));var E=e("./ops/loss_ops");n.Reduction=E.Reduction,r(e("./train")),r(e("./globals"));var _=e("./environment");n.ENV=_.ENV,n.Environment=_.Environment,n.setBackend=o.Environment.setBackend,n.getBackend=o.Environment.getBackend,n.memory=o.Environment.memory;var T=e("./doc");n.doc=T.doc,n.nextFrame=i.BrowserUtil.nextFrame,n.webgl={webgl_util:u,gpgpu_util:s}},{"./browser_util":1,"./doc":3,"./environment":5,"./globals":6,"./kernels/backend_cpu":9,"./kernels/backend_webgl":11,"./kernels/webgl/gpgpu_util":25,"./kernels/webgl/webgl_util":45,"./ops/loss_ops":59,"./ops/ops":66,"./optimizers/adadelta_optimizer":78,"./optimizers/adagrad_optimizer":79,"./optimizers/adam_optimizer":80,"./optimizers/adamax_optimizer":81,"./optimizers/momentum_optimizer":82,"./optimizers/optimizer":83,"./optimizers/rmsprop_optimizer":85,"./optimizers/sgd_optimizer":86,"./tensor":89,"./test_util":91,"./train":93,"./types":94,"./util":95,"./version":96,"./weights_loader":97}],9:[function(e,t,n){"use strict";var r=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},i=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var a=e("seedrandom"),o=e("../environment"),s=e("../ops/axis_util"),u=e("../ops/broadcast_util"),l=e("../ops/concat_util"),c=e("../ops/ops"),p=e("../ops/ops"),h=e("../ops/selu_util"),d=e("../ops/erf_util"),f=e("../tensor"),g=e("../types"),m=e("../util"),v=e("./backend_util"),y=function(){function e(){this.data=new WeakMap,"undefined"!=typeof document&&(this.canvas=document.createElement("canvas"))}return e.prototype.register=function(e,t,n){if(this.data.has(e))throw new Error("Data buffer is already registered");this.data.set(e,null)},e.prototype.write=function(e,t){if(null==t)throw new Error("MathBackendCPU.write(): values can not be null");this.throwIfNoData(e),this.data.set(e,t)},e.prototype.fromPixels=function(e,t){if(null==e)throw new Error("MathBackendCPU.writePixels(): pixels can not be null");var n;if(e instanceof ImageData)n=e.data;else if(e instanceof HTMLCanvasElement)n=e.getContext("2d").getImageData(0,0,e.width,e.height).data;else{if(!(e instanceof HTMLImageElement||e instanceof HTMLVideoElement))throw new Error("pixels is of unknown type: "+e.constructor.name);if(null==this.canvas)throw new Error("Can't read pixels from HTMLImageElement outside the browser.");this.canvas.width=e.width,this.canvas.height=e.height,this.canvas.getContext("2d").drawImage(e,0,0,e.width,e.height),n=this.canvas.getContext("2d").getImageData(0,0,e.width,e.height).data}var r;if(4===t)r=new Int32Array(n);else{var i=e.width*e.height;r=new Int32Array(i*t);for(var a=0;a<i;a++)for(var o=0;o<t;++o)r[a*t+o]=n[4*a+o]}var s=[e.height,e.width,t];return p.tensor3d(r,s,"int32")},e.prototype.read=function(e){return r(this,void 0,void 0,function(){return i(this,function(t){return[2,this.readSync(e)]})})},e.prototype.readSync=function(e){return this.throwIfNoData(e),this.data.get(e)},e.prototype.disposeData=function(e){this.data.has(e)&&this.data.delete(e)},e.prototype.time=function(e){return r(this,void 0,void 0,function(){var t,n;return i(this,function(r){return t=performance.now(),e(),n=performance.now()-t,[2,{kernelMs:n}]})})},e.prototype.memory=function(){return{unreliable:!0}},e.prototype.throwIfNoData=function(e){if(!this.data.has(e))throw new Error("CPU backend: No data found for this tensor. Did you change your backend in the middle of the program? New backends can't use Tensors created with previous backends")},e.prototype.slice=function(e,t,n){for(var r=c.buffer(n,e.dtype),i=0;i<r.size;++i){var a=r.indexToLoc(i),o=a.map(function(e,n){return e+t[n]});r.set.apply(r,[e.get.apply(e,o)].concat(a))}return r.toTensor()},e.prototype.reverse=function(e,t){for(var n=c.buffer(e.shape,e.dtype),r=e.buffer(),i=0;i<n.size;i++)!function(i){var a=n.indexToLoc(i),o=a.slice();t.forEach(function(t){return o[t]=e.shape[t]-1-o[t]}),n.set.apply(n,[r.get.apply(r,o)].concat(a))}(i);return n.toTensor()},e.prototype.concat=function(e,t){var n=l.computeOutShape(e.shape,t.shape,1),r=c.buffer(n,e.dtype);if(1===e.shape[0]&&1===t.shape[0]){var i=e.dataSync(),a=t.dataSync(),o=r.values;return o.set(i,0),o.set(a,e.size),r.toTensor()}for(var s=0;s<n[0];++s){for(u=0;u<e.shape[1];++u)r.set(e.get(s,u),s,u);for(var u=0;u<t.shape[1];++u)r.set(t.get(s,u),s,u+e.shape[1])}return r.toTensor()},e.prototype.neg=function(e){return this.multiply(c.scalar(-1),e)},e.prototype.add=function(e,t){return this.broadcastedBinaryOp(e,t,g.upcastType(e.dtype,t.dtype),function(e,t){return e+t})},e.prototype.subtract=function(e,t){return this.broadcastedBinaryOp(e,t,g.upcastType(e.dtype,t.dtype),function(e,t){return e-t})},e.prototype.pow=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){return Math.pow(e,t)})},e.prototype.matMul=function(e,t,n,r){for(var i=n?e.shape[0]:e.shape[1],a=n?e.shape[1]:e.shape[0],o=r?t.shape[0]:t.shape[1],s=e.dataSync(),u=t.dataSync(),l=n?[1,e.strides[0]]:[e.strides[0],1],p=l[0],h=l[1],d=r?[t.strides[0],1]:[1,t.strides[0]],f=d[0],g=d[1],m=a*p,v=o*f,y=new Float32Array(a*o),b=0,w=0;w<m;w+=p)for(var x=0;x<v;x+=f){for(var A=w,E=x,_=0,T=0;T<i;++T)_+=s[A]*u[E],A+=h,E+=g;y[b++]=_}return c.tensor2d(y,[a,o])},e.prototype.multiply=function(e,t){return this.broadcastedBinaryOp(e,t,g.upcastType(e.dtype,t.dtype),function(e,t){return e*t})},e.prototype.divide=function(e,t){var n,r;return"int32"===e.dtype&&"int32"===t.dtype?(r="int32",n=function(e,t){return Math.floor(e/t)}):(r="float32",n=function(e,t){return e/t}),this.broadcastedBinaryOp(e,t,r,n)},e.prototype.sum=function(e,t){s.assertAxesAreInnerMostDims("sum",t,e.rank);for(var n=s.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=g.upcastType(e.dtype,"int32"),o=c.zeros(r,a),u=m.sizeFromShape(i),l=o.dataSync(),p=e.dataSync(),h=0;h<l.length;++h){for(var d=h*u,f=0,v=0;v<u;++v)f+=p[d+v];l[h]=f}return o},e.prototype.argMin=function(e,t){var n=[t];s.assertAxesAreInnerMostDims("argMin",n,e.rank);for(var r=s.computeOutAndReduceShapes(e.shape,n),i=r[0],a=r[1],o=c.zeros(i,"int32"),u=m.sizeFromShape(a),l=o.dataSync(),p=e.dataSync(),h=0;h<l.length;++h){for(var d=h*u,f=p[d],g=0,v=0;v<u;++v){var y=p[d+v];y<f&&(f=y,g=v)}l[h]=g}return o},e.prototype.argMax=function(e,t){var n=[t];s.assertAxesAreInnerMostDims("argMax",n,e.rank);for(var r=s.computeOutAndReduceShapes(e.shape,n),i=r[0],a=r[1],o=c.zeros(i,"int32"),u=m.sizeFromShape(a),l=o.dataSync(),p=e.dataSync(),h=0;h<l.length;++h){for(var d=h*u,f=p[d],g=0,v=0;v<u;++v){var y=p[d+v];y>f&&(f=y,g=v)}l[h]=g}return o},e.prototype.equal=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e===t?1:0})},e.prototype.notEqual=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e!==t?1:0})},e.prototype.less=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e<t?1:0})},e.prototype.lessEqual=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e<=t?1:0})},e.prototype.greater=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e>t?1:0})},e.prototype.greaterEqual=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e>=t?1:0})},e.prototype.logicalNot=function(e){for(var t=e.dataSync(),n=new Int32Array(t.length),r=0;r<t.length;++r)n[r]=t[r]?0:1;return f.Tensor.make(e.shape,{values:n},"bool")},e.prototype.logicalAnd=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e&&t})},e.prototype.logicalOr=function(e,t){return this.broadcastedBinaryOp(e,t,"bool",function(e,t){return e||t})},e.prototype.where=function(e,t,n,r){for(var i=e.dataSync(),a=t.dataSync(),o=n.dataSync(),s=c.zeros(t.shape,r),u=s.dataSync(),l=0,p=0===e.rank||e.rank>1||1===t.rank?1:t.shape[1],h=0;h<i.length;h++)for(var d=0;d<p;d++)1===i[h]?u[l++]=a[h]:u[l++]=o[h];return s},e.prototype.topKValues=function(e,t){return this.topK(e,t).values},e.prototype.topKIndices=function(e,t){return this.topK(e,t).indices},e.prototype.topK=function(e,t){for(var n=e.dataSync(),r=[],i=0;i<n.length;i++)r.push({value:n[i],index:i});r.sort(function(e,t){return t.value-e.value});for(var a=m.getTypedArrayFromDType(e.dtype,t),o=new Int32Array(t),i=0;i<t;i++)a[i]=r[i].value,o[i]=r[i].index;return{values:c.tensor1d(a,e.dtype),indices:c.tensor1d(o,"int32")}},e.prototype.min=function(e,t){s.assertAxesAreInnerMostDims("min",t,e.rank);for(var n=s.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=c.zeros(r,e.dtype),o=m.sizeFromShape(i),u=a.dataSync(),l=e.dataSync(),p=0;p<u.length;++p){for(var h=p*o,d=l[0],f=0;f<o;++f){var g=l[h+f];g<d&&(d=g)}u[p]=d}return a},e.prototype.minimum=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){return Math.min(e,t)})},e.prototype.mod=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){var n=e%t;return e<0&&t<0||e>=0&&t>=0?n:(n+t)%t})},e.prototype.max=function(e,t){s.assertAxesAreInnerMostDims("max",t,e.rank);for(var n=s.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=c.zeros(r,e.dtype),o=m.sizeFromShape(i),u=a.dataSync(),l=e.dataSync(),p=0;p<u.length;++p){for(var h=p*o,d=l[h],f=0;f<o;++f){var g=l[h+f];g>d&&(d=g)}u[p]=d}return a},e.prototype.maximum=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){return Math.max(e,t)})},e.prototype.squaredDifference=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){var n=e-t;return n*n})},e.prototype.ceil=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)n[r]=Math.ceil(t[r]);return f.Tensor.make(e.shape,{values:n})},e.prototype.floor=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)n[r]=Math.floor(t[r]);return f.Tensor.make(e.shape,{values:n})},e.prototype.sign=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)t[r]<0?n[r]=-1:t[r]>0?n[r]=1:n[r]=0;return f.Tensor.make(e.shape,{values:n})},e.prototype.round=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=Math.floor(t[r]);t[r]-i<.5?n[r]=Math.floor(t[r]):t[r]-i>.5?n[r]=Math.ceil(t[r]):n[r]=i%2==0?i:i+1}return f.Tensor.make(e.shape,{values:n})},e.prototype.exp=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)n[r]=Math.exp(t[r]);return f.Tensor.make(e.shape,{values:n})},e.prototype.expm1=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)n[r]=Math.expm1(t[r]);return f.Tensor.make(e.shape,{values:n})},e.prototype.log=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=t[r];n[r]=Math.log(i)}return f.Tensor.make(e.shape,{values:n})},e.prototype.log1p=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=t[r];n[r]=Math.log1p(i)}return f.Tensor.make(e.shape,{values:n})},e.prototype.sqrt=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=t[r];n[r]=Math.sqrt(i)}return f.Tensor.make(e.shape,{values:n})},e.prototype.rsqrt=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=t[r];n[r]=1/Math.sqrt(i)}return f.Tensor.make(e.shape,{values:n})},e.prototype.square=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r){var i=t[r];n[r]=i*i}return f.Tensor.make(e.shape,{values:n})},e.prototype.reciprocal=function(e){for(var t=e.dataSync(),n=new Float32Array(t.length),r=0;r<t.length;++r)n[r]=1/t[r];return f.Tensor.make(e.shape,{values:n})},e.prototype.relu=function(e){for(var t=c.zeros(e.shape,e.dtype),n=t.dataSync(),r=e.dataSync(),i=0;i<r.length;++i)n[i]=Math.max(0,r[i]);return t},e.prototype.elu=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r){var i=n[r];t[r]=i>=0?i:Math.exp(i)-1}return f.Tensor.make(e.shape,{values:t})},e.prototype.eluDer=function(e,t){for(var n=new Float32Array(t.size),r=t.dataSync(),i=e.dataSync(),a=0;a<r.length;++a){var o=r[a];n[a]=o>=1?i[a]:i[a]*(o+1)}return f.Tensor.make(t.shape,{values:n})},e.prototype.selu=function(e){for(var t=h.SELU_SCALEALPHA,n=h.SELU_SCALE,r=new Float32Array(e.size),i=e.dataSync(),a=0;a<i.length;++a){var o=i[a];r[a]=o>=0?n*o:t*(Math.exp(o)-1)}return f.Tensor.make(e.shape,{values:r})},e.prototype.clip=function(e,t,n){for(var r=new Float32Array(e.size),i=e.dataSync(),a=0;a<i.length;++a)r[a]=Math.min(n,Math.max(t,i[a]));return f.Tensor.make(e.shape,{values:r})},e.prototype.abs=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.abs(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.int=function(e){for(var t=new Int32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=n[r];return f.Tensor.make(e.shape,{values:t},"int32")},e.prototype.sigmoid=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=1/(1+Math.exp(-n[r]));return f.Tensor.make(e.shape,{values:t})},e.prototype.softplus=function(e){for(var t=Math.log(1.1920928955078125e-7)+2,n=new Float32Array(e.size),r=e.dataSync(),i=0;i<r.length;++i){var a=r[i]>-t,o=r[i]<t,s=Math.exp(r[i]),u=void 0;u=o?s:a?r[i]:Math.log(1+s),n[i]=u}return f.Tensor.make(e.shape,{values:n})},e.prototype.sin=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.sin(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.cos=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.cos(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.tan=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.tan(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.asin=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.asin(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.acos=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.acos(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.atan=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.atan(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.atan2=function(e,t){return this.broadcastedBinaryOp(e,t,e.dtype,function(e,t){return Math.atan2(e,t)})},e.prototype.sinh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.sinh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.cosh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.cosh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.tanh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=m.tanh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.asinh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.asinh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.acosh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.acosh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.atanh=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=0;r<n.length;++r)t[r]=Math.atanh(n[r]);return f.Tensor.make(e.shape,{values:t})},e.prototype.erf=function(e){for(var t=new Float32Array(e.size),n=e.dataSync(),r=d.ERF_P,i=d.ERF_A1,a=d.ERF_A2,o=d.ERF_A3,s=d.ERF_A4,u=d.ERF_A5,l=0;l<n.length;++l){var c=n[l],p=1/(1+r*c);t[l]=1-((((u*p+s)*p+o)*p+a)*p+i)*p*Math.exp(-c*c)}return f.Tensor.make(e.shape,{values:t})},e.prototype.step=function(e,t){void 0===t&&(t=0);for(var n=new Float32Array(e.size),r=e.dataSync(),i=0;i<r.length;++i){var a=r[i];isNaN(a)?n[i]=NaN:n[i]=a>0?1:t}return f.Tensor.make(e.shape,{values:n})},e.prototype.conv2d=function(e,t,n){for(var r=n.filterHeight,i=n.filterWidth,a=n.dilationHeight,o=n.dilationWidth,s=n.padInfo.left,u=n.padInfo.top,l=c.buffer(n.outShape,e.dtype),p=0;p<n.batchSize;++p)for(var h=0;h<n.outChannels;++h)for(var d=0;d<n.outHeight;++d)for(var f=d*n.strideHeight-s,g=0;g<n.outWidth;++g){for(var m=g*n.strideWidth-u,v=0,y=0;y<r;y++){var b=f+y*a;if(!(b<0||b>=n.inHeight))for(var w=0;w<i;w++){var x=m+w*o;if(!(x<0||x>=n.inWidth))for(var A=0;A<n.inChannels;++A)v+=e.get(p,b,x,A)*t.get(y,w,A,h)}}l.set(v,p,d,g,h)}return l.toTensor()},e.prototype.conv2dDerInput=function(e,t,n){for(var r=n.filterHeight,i=n.filterWidth,a=r-1-n.padInfo.top,o=i-1-n.padInfo.left,s=n.strideHeight,u=n.strideWidth,l=c.buffer(n.inShape,"float32"),p=0;p<n.batchSize;++p)for(var h=0;h<n.inChannels;++h)for(var d=0;d<n.inHeight;++d)for(var f=d-o,g=Math.max(0,Math.ceil(f/s)),m=Math.min(n.outHeight,(r+f)/s),v=0;v<n.inWidth;++v){for(var y=v-a,b=Math.max(0,Math.ceil(y/u)),w=Math.min(n.outWidth,(i+y)/u),x=0,A=g;A<m;++A)for(var E=A*s-f,_=b;_<w;++_)for(var T=_*u-y,S=0;S<n.outChannels;++S)x+=e.get(p,A,_,S)*t.get(r-1-E,i-1-T,h,S);l.set(x,p,d,v,h)}return l.toTensor()},e.prototype.conv2dDerFilter=function(e,t,n){for(var r=n.strideHeight,i=n.strideWidth,a=n.filterHeight,o=n.filterWidth,s=c.buffer(n.filterShape,"float32"),u=n.padInfo.left,l=n.padInfo.top,p=0;p<a;++p)for(var h=Math.max(0,Math.ceil((l-p)/r)),d=Math.min(n.outHeight,(n.inHeight+l-p)/r),f=0;f<o;++f)for(var g=Math.max(0,Math.ceil((u-f)/i)),m=Math.min(n.outWidth,(n.inWidth+u-f)/i),v=0;v<n.inChannels;++v)for(var y=0;y<n.outChannels;++y){for(var b=0,w=0;w<n.batchSize;++w)for(var x=h;x<d;++x)for(var A=p+x*r-l,E=g;E<m;++E){var _=f+E*i-u;b+=e.get(w,A,_,v)*t.get(w,x,E,y)}s.set(b,p,f,v,y)}return s.toTensor()},e.prototype.depthwiseConv2D=function(e,t,n){for(var r=n.filterHeight,i=n.filterWidth,a=n.dilationHeight,o=n.dilationWidth,s=n.padInfo.left,u=n.padInfo.top,l=n.outChannels/n.inChannels,p=c.buffer(n.outShape,e.dtype),h=0;h<n.batchSize;++h)for(var d=0;d<n.inChannels;++d)for(var f=0;f<n.outHeight;++f)for(var g=f*n.strideHeight-s,m=0;m<n.outWidth;++m)for(var v=m*n.strideWidth-u,y=0;y<l;++y){for(var b=0,w=0;w<r;++w){var x=g+w*a;if(!(x<0||x>=n.inHeight))for(var A=0;A<i;++A){var E=v+A*o;E<0||E>=n.inWidth||(b+=e.get(h,x,E,d)*t.get(w,A,d,y))}}p.set(b,h,f,m,d*l+y)}return p.toTensor()},e.prototype.tile=function(e,t){for(var n=new Array(e.rank),r=0;r<n.length;r++)n[r]=e.shape[r]*t[r];for(var i=c.buffer(n,e.dtype),a=e.buffer(),r=0;r<i.values.length;++r){for(var o=i.indexToLoc(r),s=new Array(e.rank),u=0;u<s.length;u++)s[u]=o[u]%e.shape[u];var l=a.locToIndex(s);i.values[r]=a.values[l]}return i.toTensor()},e.prototype.pad=function(e,t,n){var r=t.map(function(t,n){return t[0]+e.shape[n]+t[1]}),i=t.map(function(e){return e[0]}),a=e.buffer(),o=c.buffer(r,e.dtype);0!==n&&o.values.fill(n);for(var s=0;s<e.size;s++){var u=a.indexToLoc(s),l=u.map(function(e,t){return e+i[t]});o.set.apply(o,[e.get.apply(e,u)].concat(l))}return o.toTensor()},e.prototype.transpose=function(e,t){for(var n=new Array(e.rank),r=0;r<n.length;r++)n[r]=e.shape[t[r]];for(var i=e.dataSync(),a=p.buffer(n,e.dtype),o=e.buffer(),r=0;r<e.size;++r){for(var s=o.indexToLoc(r),u=new Array(s.length),l=0;l<u.length;l++)u[l]=s[t[l]];var c=a.locToIndex(u);a.values[c]=i[r]}return a.toTensor()},e.prototype.gather=function(e,t,n){var r=e.shape.slice(),i=t.dataSync();r[n]=i.length;for(var a=p.buffer(r,e.dtype),o=e.buffer(),s=0;s<a.size;++s){var u=a.indexToLoc(s),l=u.slice();l[n]=i[u[n]];var c=o.locToIndex(l);a.values[s]=o.values[c]}return a.toTensor()},e.prototype.pool=function(e,t,n){for(var r=t.strideHeight,i=t.strideWidth,a=t.filterHeight,o=t.filterWidth,s=c.buffer(t.outShape,"float32"),u=t.padInfo.top,l=t.padInfo.left,p=0;p<t.batchSize;++p)for(var h=0;h<t.inChannels;++h)for(var d=0;d<t.outHeight;++d)for(var f=d*r-u,g=Math.max(0,f),m=Math.min(t.inHeight,a+f),v=0;v<t.outWidth;++v){for(var y=v*i-l,b=Math.max(0,y),w=Math.min(t.inWidth,o+y),x="max"===n?Number.NEGATIVE_INFINITY:Number.POSITIVE_INFINITY,A=0,E=0,_=g;_<m;++_){for(var T=b;T<w;++T){var S=e.get(p,_,T,h);"max"===n&&S>x?x=S:"avg"===n&&(A+=S,E++)}if(isNaN(x))break}s.set("avg"===n?A/E:x,p,d,v,h)}return s.toTensor()},e.prototype.maxPool=function(e,t){return this.pool(e,t,"max")},e.prototype.maxPoolPositions=function(e,t){for(var n=c.buffer(t.outShape,"int32"),r=t.strideHeight,i=t.strideWidth,a=t.filterHeight,o=t.filterWidth,s=t.padInfo.top,u=t.padInfo.left,l=0;l<t.batchSize;++l)for(var p=0;p<t.inChannels;++p)for(var h=0;h<t.outHeight;++h)for(var d=h*r-s,f=Math.max(0,d),g=Math.min(t.inHeight,a+d),m=0;m<t.outWidth;++m){for(var v=m*i-u,y=Math.max(0,v),b=Math.min(t.inWidth,o+v),w=Number.NEGATIVE_INFINITY,x=-1,A=f;A<g;++A)for(var E=A-d,_=y;_<b;++_){var T=_-v,S=e.get(l,A,_,p);S>w&&(w=S,x=E*o+T)}n.set(x,l,h,m,p)}return n.toTensor()},e.prototype.maxPoolBackprop=function(e,t,n,r){for(var i=this.maxPoolPositions(t,r),a=r.strideHeight,o=r.strideWidth,s=r.filterHeight,u=r.filterWidth,l=u-1-r.padInfo.left,p=s-1-r.padInfo.top,h=c.buffer(t.shape,"float32"),d=0;d<r.batchSize;++d)for(var f=0;f<r.inChannels;++f)for(var g=0;g<r.inHeight;++g)for(var m=0;m<r.inWidth;++m){for(var v=g-p,y=m-l,b=0,w=0;w<s;++w){var x=(v+w)/a;if(!(x<0||x>=r.outHeight||Math.floor(x)!==x))for(var A=0;A<u;++A){var E=(y+A)/o;if(!(E<0||E>=r.outWidth||Math.floor(E)!==E)){var _=s*u-1-i.get(d,x,E,f)===w*u+A?1:0;0!==_&&(b+=e.get(d,x,E,f)*_)}}}h.set(b,d,g,m,f)}return h.toTensor()},e.prototype.avgPoolBackprop=function(e,t,n){for(var r=n.strideHeight,i=n.strideWidth,a=n.filterHeight,o=n.filterWidth,s=o-1-n.padInfo.left,u=a-1-n.padInfo.top,l=c.buffer(t.shape,"float32"),p=1/(a*o),h=0;h<n.batchSize;++h)for(var d=0;d<n.inChannels;++d)for(var f=0;f<n.inHeight;++f)for(var g=0;g<n.inWidth;++g){for(var m=f-u,v=g-s,y=0,b=0;b<a;++b){var w=(m+b)/r;if(!(w<0||w>=n.outHeight||Math.floor(w)!==w))for(var x=0;x<o;++x){var A=(v+x)/i;A<0||A>=n.outWidth||Math.floor(A)!==A||(y+=e.get(h,w,A,d))}}l.set(y*p,h,f,g,d)}return l.toTensor()},e.prototype.cast=function(e,t){return v.castTensor(e,t,this)},e.prototype.reshape=function(e,t){return v.reshapeTensor(e,t)},e.prototype.avgPool=function(e,t){return this.pool(e,t,"avg").toFloat()},e.prototype.resizeBilinear=function(e,t,n,r){for(var i=e.shape,a=i[0],o=i[1],s=i[2],u=i[3],l=c.buffer([a,t,n,u],e.dtype),p=r?[o-1,s-1]:[o,s],h=r?[t-1,n-1]:[t,n],d=0;d<a;d++)for(var f=0;f<t;f++)for(var g=0;g<n;g++)for(var m=0;m<u;m++){var v=p[0]*f/h[0],y=p[1]*g/h[1],b=Math.floor(v),w=Math.min(o-1,Math.ceil(v)),x=Math.floor(y),A=Math.min(s-1,Math.ceil(y)),E=e.get(d,b,x,m),_=e.get(d,w,x,m),T=y-x,S=E+(e.get(d,b,A,m)-E)*T,C=S+(_+(e.get(d,w,A,m)-_)*T-S)*(v-b);l.set(C,d,f,g,m)}return l.toTensor()},e.prototype.resizeNearestNeighbor=function(e,t,n,r){for(var i=e.shape,a=i[0],o=i[1],s=i[2],u=i[3],l=c.buffer([a,t,n,u],e.dtype),p=r?[o-1,s-1]:[o,s],h=r?[t-1,n-1]:[t,n],d=0;d<a;d++)for(var f=0;f<t;f++)for(var g=0;g<n;g++)for(var m=0;m<u;m++){var v=p[0]*f/h[0],y=p[1]*g/h[1],b=Math.min(o-1,Math.round(v)),w=Math.min(s-1,Math.round(y)),x=e.get(d,b,w,m);l.set(x,d,f,g,m)}return l.toTensor()},e.prototype.batchNormalization=function(e,t,n,r,i,a){for(var o=e.dataSync(),s=t.dataSync(),u=n.dataSync(),l=i?i.dataSync():new Float32Array([1]),c=a?a.dataSync():new Float32Array([0]),h=new Float32Array(o.length),d=0;d<o.length;d++)h[d]=c[d%c.length]+(o[d]-s[d%s.length])*l[d%l.length]/Math.sqrt(u[d%u.length]+r);return p.tensor4d(h,e.shape)},e.prototype.localResponseNormalization4D=function(e,t,n,r,i){for(var a=c.buffer(e.shape,"float32"),o=t,s=a.shape[3]-1,u=0;u<a.shape[0];u++)for(var l=0;l<=a.shape[1];l++)for(var p=0;p<a.shape[2];p++)for(var h=0;h<a.shape[3];h++){var d=function(t,n,r,i){for(var a=0,u=Math.max(0,i-o);u<=Math.min(i+o,s);u++){var l=e.get(t,n,r,u);a+=l*l}return a}(u,l,p,h),f=e.get(u,l,p,h)*Math.pow(n+r*d,-i);a.set(f,u,l,p,h)}return a.toTensor()},e.prototype.multinomial=function(e,t,n,r){for(var i=t?e:c.softmax(e),o=i.shape[0],s=i.shape[1],u=c.zeros([o,n],"int32"),l=u.dataSync(),p=i.dataSync(),h=0;h<o;++h){var d=h*s,f=new Float32Array(s-1);f[0]=p[d];for(var g=1;g<f.length;++g)f[g]=f[g-1]+p[d+g];for(var m=a.alea(r.toString()),v=h*n,y=0;y<n;++y){var b=m();l[v+y]=f.length;for(var w=0;w<f.length;w++)if(b<f[w]){l[v+y]=w;break}}}return u},e.prototype.oneHot=function(e,t,n,r){var i=new Float32Array(e.size*t);i.fill(r);for(var a=0;a<e.size;++a)i[a*t+e.get(a)]=n;return c.tensor2d(i,[e.size,t])},e.prototype.broadcastedBinaryOp=function(e,t,n,r){for(var i=u.assertAndGetBroadcastShape(e.shape,t.shape),a=c.buffer(i,n),o=e.dataSync(),s=t.dataSync(),l=u.getBroadcastDims(e.shape,i),p=u.getBroadcastDims(t.shape,i),h=e.buffer(),d=t.buffer(),f=0;f<a.values.length;++f)!function(n){var i=a.indexToLoc(n),u=i.slice(-e.rank);l.forEach(function(e){return u[e]=0});var c=h.locToIndex(u),f=i.slice(-t.rank);p.forEach(function(e){return f[e]=0});var g=d.locToIndex(f);a.values[n]=r(o[c],s[g])}(f);return a.toTensor()},e.prototype.dispose=function(){},e}();n.MathBackendCPU=y,o.ENV.registerBackend("cpu",function(){return new y},1)},{"../environment":5,"../ops/axis_util":47,"../ops/broadcast_util":50,"../ops/concat_util":53,"../ops/erf_util":56,"../ops/ops":66,"../ops/selu_util":72,"../tensor":89,"../types":94,"../util":95,"./backend_util":10,seedrandom:135}],10:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e(".."),i=e("../ops/array_ops");n.castTensor=function(e,t,n){if(!r.util.hasEncodingLoss(e.dtype,t))return r.Tensor.make(e.shape,{dataId:e.dataId},t);if("int32"===t)return n.int(e);if("bool"===t)return n.notEqual(e,i.ArrayOps.scalar(0,e.dtype));throw new Error("Error in Cast: unknown dtype argument ("+t+")")},n.reshapeTensor=function(e,t){return r.Tensor.make(t,{dataId:e.dataId},e.dtype)}},{"..":8,"../ops/array_ops":46}],11:[function(e,t,n){"use strict";function r(e,t){if("float32"===t)return e;if("int32"===t||"bool"===t){for(var n="int32"===t?new Int32Array(e.length):new Uint8Array(e.length),r=0;r<n.length;++r)n[r]=Math.round(e[r]);return n}throw new Error("Unknown dtype "+t)}function i(e,t){return e instanceof Float32Array?e:new Float32Array(e)}var a=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},o=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var s=e("../environment"),u=e("../ops/axis_util"),l=e("../ops/ops"),c=e("../ops/reduce_util"),p=e("../tensor"),h=e("../types"),d=e("../util"),f=e("./backend_util"),g=e("./webgl/argminmax_gpu"),m=e("./webgl/avg_pool_backprop_gpu"),v=e("./webgl/batchnorm_gpu"),y=e("./webgl/binaryop_gpu"),b=e("./webgl/binaryop_gpu"),w=e("./webgl/clip_gpu"),x=e("./webgl/concat_gpu"),A=e("./webgl/conv_backprop_gpu"),E=e("./webgl/conv_gpu"),_=e("./webgl/conv_gpu_depthwise"),T=e("./webgl/from_pixels_gpu"),S=e("./webgl/gather_gpu"),C=e("./webgl/gpgpu_context"),O=e("./webgl/gpgpu_math"),k=e("./webgl/logical_gpu"),R=e("./webgl/lrn_gpu"),N=e("./webgl/max_pool_backprop_gpu"),I=e("./webgl/mulmat_gpu"),z=e("./webgl/multinomial_gpu"),M=e("./webgl/onehot_gpu"),D=e("./webgl/pad_gpu"),P=e("./webgl/pool_gpu"),L=e("./webgl/reduce_gpu"),F=e("./webgl/resize_bilinear_gpu"),V=e("./webgl/resize_nearest_neighbor_gpu"),B=e("./webgl/reverse_gpu"),U=e("./webgl/slice_gpu"),j=e("./webgl/tex_util"),G=e("./webgl/texture_manager"),W=e("./webgl/tile_gpu"),q=e("./webgl/transpose_gpu"),H=e("./webgl/unaryop_gpu"),K=e("./webgl/unaryop_gpu"),X=e("./webgl/webgl_util"),Y=function(){function e(e,t){if(void 0===t&&(t=!0),this.gpgpu=e,this.delayedStorage=t,this.texData=new WeakMap,this.uploadWaitMs=0,this.downloadWaitMs=0,this.binaryCache={},this.disposed=!1,s.ENV.get("WEBGL_VERSION")<1)throw new Error("WebGL is not supported on this device");null==e?(this.gpgpu=new C.GPGPUContext,this.gpgpuCreatedLocally=!0):this.gpgpuCreatedLocally=!1,"undefined"!=typeof document&&(this.canvas=document.createElement("canvas")),this.textureManager=new G.TextureManager(this.gpgpu)}return e.prototype.register=function(e,t,n){if(this.texData.has(e))throw new Error("Data buffer is already registered");this.texData.set(e,{shape:t,dtype:n,values:null,texture:null,texShape:null,texType:j.TextureType.FLOAT})},e.prototype.fromPixels=function(e,t){if(null==e)throw new Error("MathBackendWebGL.writePixels(): pixels can not be null");var n=[e.height,e.width],r=[e.height,e.width,t];if(e instanceof HTMLVideoElement){if(null==this.canvas)throw new Error("Can't read pixels from HTMLImageElement outside the browser.");this.canvas.width=e.width,this.canvas.height=e.height,this.canvas.getContext("2d").drawImage(e,0,0,e.width,e.height),e=this.canvas}var i=p.Tensor.make(n,{},"int32");this.texData.get(i.dataId).texType=j.TextureType.UNSIGNED_BYTE,this.gpgpu.uploadPixelDataToTexture(this.getTexture(i.dataId),e);var a=new T.FromPixelsProgram(r),o=this.compileAndRun(a,[i]);return i.dispose(),o},e.prototype.write=function(e,t){if(null==t)throw new Error("MathBackendWebGL.write(): values can not be null");this.throwIfNoData(e);var n=this.texData.get(e),r=n.texture,i=n.texShape,a=n.texType;null!=r&&(this.textureManager.releaseTexture(r,i,a),n.texture=null,n.texShape=null),n.values=t,this.delayedStorage||this.uploadToGPU(e)},e.prototype.readSync=function(e){this.throwIfNoData(e);var t=this.texData.get(e),n=t.texture,r=t.values,i=t.texShape;if(null!=r)return this.cacheOnCPU(e),r;var a,o=null!=this.activeTimers;o&&(a=performance.now());var s=this.gpgpu.downloadMatrixFromTexture(n,i[0],i[1]);return o&&(this.downloadWaitMs+=performance.now()-a),this.cacheOnCPU(e,s),t.values},e.prototype.read=function(e){return a(this,void 0,void 0,function(){var t,n,r,i,a;return o(this,function(o){switch(o.label){case 0:return this.throwIfNoData(e),t=this.texData.get(e),n=t.texture,r=t.values,i=t.texShape,null!=r?(this.cacheOnCPU(e),[2,r]):s.ENV.get("WEBGL_GET_BUFFER_SUB_DATA_ASYNC_EXTENSION_ENABLED")?[4,this.gpgpu.downloadMatrixFromTextureAsync(n,i[0],i[1])]:[3,2];case 1:return a=o.sent(),this.cacheOnCPU(e,a),[2,t.values];case 2:return 0===s.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")?[2,this.readSync(e)]:[4,this.gpgpu.runQuery(function(){})];case 3:return o.sent(),[2,this.readSync(e)]}})})},e.prototype.time=function(e){return a(this,void 0,void 0,function(){var t,n,r,i,a,s;return o(this,function(o){switch(o.label){case 0:return t=this.activeTimers,n=[],r=!1,null==this.programTimersStack?(this.programTimersStack=n,r=!0):this.activeTimers.push(n),this.activeTimers=n,e(),i=d.flatten(this.activeTimers),this.activeTimers=t,r&&(this.programTimersStack=null),[4,Promise.all(i).then(function(e){var t=0;return e.forEach(function(e){return t+=e}),t})];case 1:return a=o.sent(),s={uploadWaitMs:this.uploadWaitMs,downloadWaitMs:this.downloadWaitMs,kernelMs:a,wallMs:null},this.uploadWaitMs=0,this.downloadWaitMs=0,[2,s]}})})},e.prototype.memory=function(){return{unreliable:!1}},e.prototype.startTimer=function(){return s.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")>0?this.gpgpu.beginQuery():{startMs:performance.now(),endMs:null}},e.prototype.endTimer=function(e){return s.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")>0?(this.gpgpu.endQuery(),e):(e.endMs=performance.now(),e)},e.prototype.getQueryTime=function(e){return a(this,void 0,void 0,function(){var t;return o(this,function(n){return s.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")>0?[2,this.gpgpu.pollQueryTime(e)]:(t=e,[2,t.endMs-t.startMs])})})},e.prototype.disposeData=function(e){if(this.texData.has(e)){var t=this.texData.get(e),n=t.texture,r=t.texShape,i=t.texType;null!=n&&this.textureManager.releaseTexture(n,r,i),this.texData.delete(e)}},e.prototype.getTexture=function(e){return this.uploadToGPU(e),this.texData.get(e).texture},e.prototype.getTextureData=function(e){return this.uploadToGPU(e),this.texData.get(e)},e.prototype.getGPGPUContext=function(){return this.gpgpu},e.prototype.slice=function(e,t,n){var r=new U.SliceProgram(n),i=r.getCustomSetupFunc(t);return this.compileAndRun(r,[e],null,i)},e.prototype.reverse=function(e,t){var n=new B.ReverseProgram(e.shape,t);return this.compileAndRun(n,[e])},e.prototype.concat=function(e,t){var n=new x.ConcatProgram(e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.neg=function(e){var t=new K.UnaryOpProgram(e.shape,H.NEG);return this.compileAndRun(t,[e])},e.prototype.matMul=function(e,t,n,r){var i=new I.MatMulProgram(e.shape,t.shape,n,r);return this.compileAndRun(i,[e,t])},e.prototype.multiply=function(e,t){var n=new b.BinaryOpProgram(y.MUL,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,h.upcastType(e.dtype,t.dtype));return this.compileAndRun(n,[e,t],r)},e.prototype.batchNormalization=function(e,t,n,r,i,a){var o=[e,t,n],s=null;null!=a&&(s=a.shape,o.push(a));var u=null;null!=i&&(u=i.shape,o.push(i));var l=new v.BatchNormProgram(e.shape,t.shape,n.shape,s,u,r);return this.compileAndRun(l,o)},e.prototype.localResponseNormalization4D=function(e,t,n,r,i){var a=new R.LRNProgram(e.shape,t,n,r,i);return this.compileAndRun(a,[e])},e.prototype.tile=function(e,t){var n=new W.TileProgram(e.shape,t);return this.compileAndRun(n,[e])},e.prototype.pad=function(e,t,n){var r=new D.PadProgram(e.shape,t,n);return this.compileAndRun(r,[e])},e.prototype.transpose=function(e,t){var n=new q.TransposeProgram(e.shape,t);return this.compileAndRun(n,[e])},e.prototype.gather=function(e,t,n){var r=new S.GatherProgram(e.shape,t.size,n);return this.compileAndRun(r,[e,t])},e.prototype.reduce=function(e,t,n){var r=e.shape[0],i=e.shape[1],a={windowSize:c.computeOptimalWindowSize(i),inSize:i,batchSize:r},o=new L.ReduceProgram(a,t),s=o.outputShape,u=s[0],l=s[1],p=this.makeOutputArray([u,l],n);return this.compileAndRun(o,[e],p),1===p.shape[1]?p:this.reduce(p,t,n)},e.prototype.argReduce=function(e,t,n){void 0===n&&(n=null);var r=e.shape[0],i=e.shape[1];null!=n&&(r=n.shape[0],i=n.shape[1]);var a={windowSize:c.computeOptimalWindowSize(i),inSize:i,batchSize:r},o=new g.ArgMinMaxProgram(a,t,null==n),s=o.outputShape,u=s[0],l=s[1],p=this.makeOutputArray([u,l],"int32"),h=[e];return null!=n&&h.push(n),this.compileAndRun(o,h,p),1===p.shape[1]?p:this.argReduce(e,t,p)},e.prototype.sum=function(e,t){u.assertAxesAreInnerMostDims("sum",t,e.rank);var n=u.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=d.sizeFromShape(i),o=e.as2D(-1,a),s=h.sumOutType(e.dtype);return this.reduce(o,"sum",s).reshape(r)},e.prototype.argMin=function(e,t){var n=[t];u.assertAxesAreInnerMostDims("argMin",n,e.rank);var r=u.computeOutAndReduceShapes(e.shape,n),i=r[0],a=r[1],o=d.sizeFromShape(a),s=e.as2D(-1,o);return this.argReduce(s,"min").reshape(i)},e.prototype.argMax=function(e,t){var n=[t];u.assertAxesAreInnerMostDims("argMax",n,e.rank);var r=u.computeOutAndReduceShapes(e.shape,n),i=r[0],a=r[1],o=d.sizeFromShape(a),s=e.as2D(-1,o);return this.argReduce(s,"max").reshape(i)},e.prototype.equal=function(e,t){var n=new b.BinaryOpProgram(y.EQUAL,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.notEqual=function(e,t){var n=new b.BinaryOpProgram(y.NOT_EQUAL,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.less=function(e,t){var n=new b.BinaryOpProgram(y.LESS,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.lessEqual=function(e,t){var n=new b.BinaryOpProgram(y.LESS_EQUAL,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.greater=function(e,t){var n=new b.BinaryOpProgram(y.GREATER,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.greaterEqual=function(e,t){var n=new b.BinaryOpProgram(y.GREATER_EQUAL,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.logicalNot=function(e){var t=new K.UnaryOpProgram(e.shape,H.LOGICAL_NOT);return this.compileAndRun(t,[e])},e.prototype.logicalAnd=function(e,t){var n=new b.BinaryOpProgram(y.LOGICAL_AND,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.logicalOr=function(e,t){var n=new b.BinaryOpProgram(y.LOGICAL_OR,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,"bool");return this.compileAndRun(n,[e,t],r)},e.prototype.where=function(e,t,n,r){var i=new k.WhereProgram(e.rank,t.shape,t.rank),a=this.makeOutputArray(i.outputShape,r);return this.compileAndRun(i,[e,t,n],a)},e.prototype.topKValues=function(e,t){throw new Error("topKValues GPU not yet implemented!")},e.prototype.topKIndices=function(e,t){throw new Error("topKIndices GPU not yet implemented!")},e.prototype.min=function(e,t){u.assertAxesAreInnerMostDims("min",t,e.rank);var n=u.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=d.sizeFromShape(i),o=e.as2D(-1,a);return this.reduce(o,"min",o.dtype).reshape(r)},e.prototype.minimum=function(e,t){var n=new b.BinaryOpProgram(y.MIN,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.mod=function(e,t){var n=new b.BinaryOpProgram(y.MOD,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.max=function(e,t){u.assertAxesAreInnerMostDims("max",t,e.rank);var n=u.computeOutAndReduceShapes(e.shape,t),r=n[0],i=n[1],a=d.sizeFromShape(i),o=e.as2D(-1,a);return this.reduce(o,"max",o.dtype).reshape(r)},e.prototype.maximum=function(e,t){var n=new b.BinaryOpProgram(y.MAX,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.squaredDifference=function(e,t){var n=new b.BinaryOpProgram(y.SQUARED_DIFFERENCE,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.divide=function(e,t){var n,r;"int32"===e.dtype&&"int32"===t.dtype?(n=y.INT_DIV,r="int32"):(n=y.DIV,r="float32");var i=new b.BinaryOpProgram(n,e.shape,t.shape),a=this.makeOutputArray(i.outputShape,r);return this.compileAndRun(i,[e,t],a)},e.prototype.add=function(e,t){var n=new b.BinaryOpProgram(y.ADD,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,h.upcastType(e.dtype,t.dtype));return this.compileAndRun(n,[e,t],r)},e.prototype.subtract=function(e,t){var n=new b.BinaryOpProgram(y.SUB,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,h.upcastType(e.dtype,t.dtype));return this.compileAndRun(n,[e,t],r)},e.prototype.pow=function(e,t){var n=new b.BinaryOpProgram(y.POW,e.shape,t.shape),r=this.makeOutputArray(n.outputShape,h.upcastType(e.dtype,t.dtype));return this.compileAndRun(n,[e,t],r)},e.prototype.ceil=function(e){var t=new K.UnaryOpProgram(e.shape,H.CEIL);return this.compileAndRun(t,[e])},e.prototype.floor=function(e){var t=new K.UnaryOpProgram(e.shape,H.FLOOR);return this.compileAndRun(t,[e])},e.prototype.sign=function(e){var t=new K.UnaryOpProgram(e.shape,H.SIGN);return this.compileAndRun(t,[e])},e.prototype.round=function(e){var t=new K.UnaryOpProgram(e.shape,H.ROUND);return this.compileAndRun(t,[e])},e.prototype.exp=function(e){var t=new K.UnaryOpProgram(e.shape,H.EXP);return this.compileAndRun(t,[e])},e.prototype.expm1=function(e){var t=new K.UnaryOpProgram(e.shape,H.EXPM1);return this.compileAndRun(t,[e])},e.prototype.log=function(e){var t=new K.UnaryOpProgram(e.shape,H.LOG);return this.compileAndRun(t,[e])},e.prototype.log1p=function(e){var t=new K.UnaryOpProgram(e.shape,H.LOG1P);return this.compileAndRun(t,[e])},e.prototype.sqrt=function(e){var t=new K.UnaryOpProgram(e.shape,H.SQRT);return this.compileAndRun(t,[e])},e.prototype.rsqrt=function(e){var t=new K.UnaryOpProgram(e.shape,H.RSQRT);return this.compileAndRun(t,[e])},e.prototype.square=function(e){var t=new K.UnaryOpProgram(e.shape,H.SQUARE);return this.compileAndRun(t,[e])},e.prototype.reciprocal=function(e){var t=new K.UnaryOpProgram(e.shape,H.RECIPROCAL);return this.compileAndRun(t,[e])},e.prototype.relu=function(e){var t=new K.UnaryOpProgram(e.shape,H.RELU);return this.compileAndRun(t,[e])},e.prototype.elu=function(e){var t=new K.UnaryOpProgram(e.shape,H.ELU);return this.compileAndRun(t,[e])},e.prototype.eluDer=function(e,t){var n=new b.BinaryOpProgram(y.ELU_DER,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.selu=function(e){var t=new K.UnaryOpProgram(e.shape,H.SELU);return this.compileAndRun(t,[e])},e.prototype.int=function(e){var t=new K.UnaryOpProgram(e.shape,H.TO_INT),n=this.makeOutputArray(t.outputShape,"int32");return this.compileAndRun(t,[e],n)},e.prototype.clip=function(e,t,n){var r=new w.ClipProgram(e.shape,t,n);return this.compileAndRun(r,[e])},e.prototype.abs=function(e){var t=new K.UnaryOpProgram(e.shape,H.ABS);return this.compileAndRun(t,[e])},e.prototype.sigmoid=function(e){var t=new K.UnaryOpProgram(e.shape,H.SIGMOID);return this.compileAndRun(t,[e])},e.prototype.softplus=function(e){var t=new K.UnaryOpProgram(e.shape,H.SOFTPLUS);return this.compileAndRun(t,[e])},e.prototype.sin=function(e){var t=new K.UnaryOpProgram(e.shape,H.SIN);return this.compileAndRun(t,[e])},e.prototype.cos=function(e){var t=new K.UnaryOpProgram(e.shape,H.COS);return this.compileAndRun(t,[e])},e.prototype.tan=function(e){var t=new K.UnaryOpProgram(e.shape,H.TAN);return this.compileAndRun(t,[e])},e.prototype.asin=function(e){var t=new K.UnaryOpProgram(e.shape,H.ASIN);return this.compileAndRun(t,[e])},e.prototype.acos=function(e){var t=new K.UnaryOpProgram(e.shape,H.ACOS);return this.compileAndRun(t,[e])},e.prototype.atan=function(e){var t=new K.UnaryOpProgram(e.shape,H.ATAN);return this.compileAndRun(t,[e])},e.prototype.atan2=function(e,t){var n=new b.BinaryOpProgram(y.ATAN2,e.shape,t.shape);return this.compileAndRun(n,[e,t])},e.prototype.sinh=function(e){var t=new K.UnaryOpProgram(e.shape,H.SINH);return this.compileAndRun(t,[e])},e.prototype.cosh=function(e){var t=new K.UnaryOpProgram(e.shape,H.COSH);return this.compileAndRun(t,[e])},e.prototype.tanh=function(e){var t=new K.UnaryOpProgram(e.shape,H.TANH);return this.compileAndRun(t,[e])},e.prototype.asinh=function(e){var t=new K.UnaryOpProgram(e.shape,H.ASINH);return this.compileAndRun(t,[e])},e.prototype.acosh=function(e){var t=new K.UnaryOpProgram(e.shape,H.ACOSH);return this.compileAndRun(t,[e])},e.prototype.atanh=function(e){var t=new K.UnaryOpProgram(e.shape,H.ATANH);return this.compileAndRun(t,[e])},e.prototype.erf=function(e){var t=new K.UnaryOpProgram(e.shape,H.ERF);return this.compileAndRun(t,[e])},e.prototype.step=function(e,t){var n=new K.UnaryOpProgram(e.shape,H.STEP(t));return this.compileAndRun(n,[e])},e.prototype.conv2d=function(e,t,n){var r=new E.Conv2DProgram(n);return this.compileAndRun(r,[e,t])},e.prototype.conv2dDerInput=function(e,t,n){var r=new A.Conv2DDerInputProgram(n);return this.compileAndRun(r,[e,t])},e.prototype.conv2dDerFilter=function(e,t,n){var r=new A.Conv2DDerFilterProgram(n);return this.compileAndRun(r,[e,t])},e.prototype.depthwiseConv2D=function(e,t,n){var r=new _.DepthwiseConv2DProgram(n);return this.compileAndRun(r,[e,t])},e.prototype.maxPool=function(e,t){var n=new P.Pool2DProgram(t,"max",!1),r=this.makeOutputArray(n.outputShape,e.dtype);return this.compileAndRun(n,[e],r)},e.prototype.avgPool=function(e,t){var n=new P.Pool2DProgram(t,"avg",!1),r=this.makeOutputArray(n.outputShape,"float32");return this.compileAndRun(n,[e],r)},e.prototype.maxPoolBackprop=function(e,t,n,r){var i=new P.Pool2DProgram(r,"max",!0),a=this.compileAndRun(i,[t]),o=new N.MaxPool2DBackpropProgram(r),s=this.makeOutputArray(o.outputShape,t.dtype),u=this.compileAndRun(o,[e,a],s);return a.dispose(),u},e.prototype.avgPoolBackprop=function(e,t,n){var r=new m.AvgPool2DBackpropProgram(n),i=this.makeOutputArray(r.outputShape,t.dtype);return this.compileAndRun(r,[e],i)},e.prototype.cast=function(e,t){return f.castTensor(e,t,this)},e.prototype.reshape=function(e,t){return f.reshapeTensor(e,t)},e.prototype.resizeBilinear=function(e,t,n,r){var i=new F.ResizeBilinearProgram(e.shape,t,n,r);return this.compileAndRun(i,[e])},e.prototype.resizeNearestNeighbor=function(e,t,n,r){var i=new V.ResizeNearestNeighborProgram(e.shape,t,n,r);return this.compileAndRun(i,[e])},e.prototype.multinomial=function(e,t,n,r){var i=t?e:l.softmax(e),a=i.shape[0],o=i.shape[1],s=new z.MultinomialProgram(a,o,n),u=this.makeOutputArray(s.outputShape,"int32"),c=s.getCustomSetupFunc(r);return this.compileAndRun(s,[i],u,c)},e.prototype.oneHot=function(e,t,n,r){var i=new M.OneHotProgram(e.size,t,n,r);return this.compileAndRun(i,[e])},e.prototype.makeOutputArray=function(e,t){return p.Tensor.make(e,{},t)},e.prototype.compileAndRun=function(e,t,n,r){var i=this;null==n&&(n=this.makeOutputArray(e.outputShape,t[0].dtype));var a=t.map(function(e){return i.uploadToGPU(e.dataId),{tensor:e,texData:i.texData.get(e.dataId)}});this.uploadToGPU(n.dataId);var o,s={tensor:n,texData:this.texData.get(n.dataId)},u=O.makeShaderKey(e,a,s),l=this.getAndSaveBinary(u,function(){return O.compileProgram(i.gpgpu,e,a,s)}),c=null!=this.activeTimers;return c&&(o=this.startTimer()),O.runProgram(l,a,s,r),c&&(o=this.endTimer(o),this.activeTimers.push(this.getQueryTime(o))),n},e.prototype.getAndSaveBinary=function(e,t){return e in this.binaryCache||(this.binaryCache[e]=t()),this.binaryCache[e]},e.prototype.getTextureManager=function(){return this.textureManager},e.prototype.dispose=function(){if(!this.disposed){for(var e in this.binaryCache)this.gpgpu.deleteProgram(this.binaryCache[e].webGLProgram);this.textureManager.dispose(),this.canvas.remove(),this.gpgpuCreatedLocally&&this.gpgpu.dispose(),this.disposed=!0}},e.prototype.throwIfNoData=function(e){if(!this.texData.has(e))throw new Error("WebGL backend: No data found for this tensor. Did you change your backend in the middle of the program? New backends can't use Tensors created with previous backends")},e.prototype.uploadToGPU=function(e){this.throwIfNoData(e);var t=this.texData.get(e),n=t.shape,r=t.values,a=t.texture,o=(t.dtype,t.texType);if(null==a){var s,u=null!=this.activeTimers;u&&(s=performance.now());var l=X.getTextureShapeFromLogicalShape(this.gpgpu.gl,n);t.texShape=l;var c=this.textureManager.acquireTexture(l,o);t.texture=c,null!=r&&(this.gpgpu.uploadMatrixToTexture(c,l[0],l[1],i(r)),t.values=null,u&&(this.uploadWaitMs+=performance.now()-s))}},e.prototype.cacheOnCPU=function(e,t){var n=this.delayedStorage,i=this.texData.get(e),a=i.texture,o=i.texShape,s=i.dtype,u=i.texType;n&&null!=a&&(this.textureManager.releaseTexture(a,o,u),i.texture=null,i.texShape=null),null!=t&&(i.values=r(t,s))},e}();n.MathBackendWebGL=Y,s.ENV.registerBackend("webgl",function(){return new Y},2)},{"../environment":5,"../ops/axis_util":47,"../ops/ops":66,"../ops/reduce_util":69,"../tensor":89,"../types":94,"../util":95,"./backend_util":10,"./webgl/argminmax_gpu":12,"./webgl/avg_pool_backprop_gpu":13,"./webgl/batchnorm_gpu":14,"./webgl/binaryop_gpu":15,"./webgl/clip_gpu":16,"./webgl/concat_gpu":17,"./webgl/conv_backprop_gpu":18,"./webgl/conv_gpu":19,"./webgl/conv_gpu_depthwise":20,"./webgl/from_pixels_gpu":21,"./webgl/gather_gpu":22,"./webgl/gpgpu_context":23,"./webgl/gpgpu_math":24,"./webgl/logical_gpu":26,"./webgl/lrn_gpu":27,"./webgl/max_pool_backprop_gpu":28,"./webgl/mulmat_gpu":29,"./webgl/multinomial_gpu":30,"./webgl/onehot_gpu":31,"./webgl/pad_gpu":32,"./webgl/pool_gpu":33,"./webgl/reduce_gpu":34,"./webgl/resize_bilinear_gpu":35,"./webgl/resize_nearest_neighbor_gpu":36,"./webgl/reverse_gpu":37,"./webgl/slice_gpu":39,"./webgl/tex_util":40,"./webgl/texture_manager":41,"./webgl/tile_gpu":42,"./webgl/transpose_gpu":43,"./webgl/unaryop_gpu":44,"./webgl/webgl_util":45}],12:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n){this.variableNames=["A"];var r=e.windowSize,i=e.batchSize,a=e.inSize,o=Math.ceil(a/r);n||this.variableNames.push("bestIndicesA"),this.outputShape=[i,o];var s="max"===t?">":"<",u=n?"inOffset + i;":"round(getBestIndicesA(batch, inOffset + i));";this.userCode="\n      void main() {\n        ivec2 coords = getOutputCoords();\n        int batch = coords[0];\n        int outIdx = coords[1];\n        int inOffset = outIdx * "+r+";\n\n        int bestIndex = 0;\n        float bestValue = getA(batch, inOffset);\n\n        for (int i = 0; i < "+r+"; i++) {\n          int inIdx = "+u+";\n          float candidate = getA(batch, inIdx);\n          if (candidate "+s+" bestValue) {\n            bestValue = candidate;\n            bestIndex = inIdx;\n          }\n        }\n        setOutput(float(bestIndex));\n      }\n    "}}();n.ArgMinMaxProgram=r},{}],13:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["dy"],this.outputShape=e.inShape;var t=e.filterHeight,n=e.filterWidth,r=e.strideHeight,i=e.strideWidth,a=t-1-e.padInfo.top,o=n-1-e.padInfo.left,s=1/(t*n);this.userCode="\n      const ivec2 pads = ivec2("+a+", "+o+");\n      const float avgMultiplier = float("+s+");\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int b = coords[0];\n        int d = coords[3];\n\n        ivec2 dyRCCorner = coords.yz - pads;\n        int dyRCorner = dyRCCorner.x;\n        int dyCCorner = dyRCCorner.y;\n\n        // Convolve dy(?, ?, d) with pos mask(:, :, d) to get dx(xR, xC, d).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n        for (int wR = 0; wR < "+t+"; wR++) {\n          float dyR = float(dyRCorner + wR) / "+r+".0;\n\n          if (dyR < 0.0 || dyR >= "+e.outHeight+".0 || fract(dyR) > 0.0) {\n            continue;\n          }\n          int idyR = int(dyR);\n\n          for (int wC = 0; wC < "+n+"; wC++) {\n            float dyC = float(dyCCorner + wC) / "+i+".0;\n\n            if (dyC < 0.0 || dyC >= "+e.outWidth+".0 ||\n                fract(dyC) > 0.0) {\n              continue;\n            }\n            int idyC = int(dyC);\n\n            float dyValue = getDy(b, idyR, idyC, d);\n\n            dotProd += dyValue * avgMultiplier;\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.AvgPool2DBackpropProgram=r},{}],14:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../../ops/broadcast_util"),i=function(){return function(e,t,n,i,a,o){this.outputShape=[],this.supportsBroadcasting=!0,this.variableNames=["x","mean","variance"],r.assertAndGetBroadcastShape(e,t),r.assertAndGetBroadcastShape(e,n);var s="0.0";null!=i&&(r.assertAndGetBroadcastShape(e,i),this.variableNames.push("offset"),s="getOffsetAtOutCoords()");var u="1.0";null!=a&&(r.assertAndGetBroadcastShape(e,a),this.variableNames.push("scale"),u="getScaleAtOutCoords()"),this.outputShape=e,this.userCode="\n      void main() {\n        float x = getXAtOutCoords();\n        float mean = getMeanAtOutCoords();\n        float variance = getVarianceAtOutCoords();\n        float offset = "+s+";\n        float scale = "+u+";\n        float inv = scale * inversesqrt(variance + float("+o+"));\n        setOutput((x - mean) * inv + offset);\n      }\n    "}}();n.BatchNormProgram=i},{"../../ops/broadcast_util":50}],15:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../../ops/broadcast_util"),i="\n  if (isNaN(a)) return a;\n  if (isNaN(b)) return b;\n";n.ADD="return a + b;",n.SUB="return a - b;",n.MUL="return a * b;",n.DIV="return a / b;",n.INT_DIV="\n  float resultSign = sign(a) * sign(b);\n  int ia = round(a);\n  int ib = round(b);\n  int result = ia / ib;\n  int amodb = ia - ib * result;\n\n  if (resultSign < 0.0 && amodb != 0) {\n    result -= 1;\n  }\n  return float(result);\n",n.POW="\n  return (round(mod(b, 2.0)) == 0 || round(mod(b, 2.0)) == 2) ?\n      pow(abs(a), b) : sign(a) * pow(abs(a), b);\n",n.SQUARED_DIFFERENCE="return (a - b) * (a - b);",n.EQUAL="return float(a == b);",n.NOT_EQUAL="return float(a != b);",n.LESS="return float(a < b);",n.LESS_EQUAL="return float(a <= b);",n.GREATER="return float(a > b);",n.GREATER_EQUAL="return float(a >= b);",n.LOGICAL_AND="return float(a >= 1.0 && b >= 1.0);",n.LOGICAL_OR="return float(a >= 1.0 || b >= 1.0);",n.MAX=i+"\n  return max(a, b);\n",n.MIN=i+"\n  return min(a, b);\n",n.MOD="return mod(a, b);",n.ATAN2=i+"\n  return atan(a, b);\n",n.ELU_DER="return (b >= 1.0) ? a : a * (b + 1.0);";var a=function(){return function(e,t,n){this.variableNames=["A","B"],this.supportsBroadcasting=!0,this.outputShape=r.assertAndGetBroadcastShape(t,n),this.userCode="\n      float binaryOperation(float a, float b) {\n        "+e+"\n      }\n\n      void main() {\n        float a = getAAtOutCoords();\n        float b = getBAtOutCoords();\n        setOutput(binaryOperation(a, b));\n      }\n    "}}();n.BinaryOpProgram=a},{"../../ops/broadcast_util":50}],16:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n){this.variableNames=["A"],this.outputShape=e;var r=t.toFixed(20),i=n.toFixed(20);this.userCode="\n      void main() {\n        float value = getAAtOutCoords();\n        if (isNaN(value)) {\n          setOutput(value);\n          return;\n        }\n\n        setOutput(clamp(value, "+r+", "+i+"));\n      }\n    "}}();n.ClipProgram=r},{}],17:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../../ops/concat_util"),i=function(){return function(e,t){this.variableNames=["A","B"],this.outputShape=[],this.outputShape=r.computeOutShape(e,t,1),this.userCode="\n      void main() {\n        ivec2 coords = getOutputCoords();\n        int yR = coords.x;\n        int yC = coords.y;\n\n        float value = 0.0;\n        if (yC < "+e[1]+") {\n          value = getA(yR, yC);\n        } else {\n          yC -= "+e[1]+";\n          value = getB(yR, yC);\n        }\n\n        setOutput(value);\n      }\n    "}}();n.ConcatProgram=i},{"../../ops/concat_util":53}],18:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["x","dy"],this.outputShape=e.filterShape;var t=e.strideHeight,n=e.strideWidth,r=e.padInfo.top,i=e.padInfo.left;this.userCode="\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int wR = coords.x;\n        int wC = coords.y;\n        int d1 = coords.z;\n        int d2 = coords.w;\n\n        // Convolve x(?, ?, d1) with dy(:, :, d2) to get dw(wR, wC, d1, d2).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n\n        for (int b = 0; b < "+e.batchSize+"; b++) {\n          for (int yR = 0; yR < "+e.outHeight+"; yR++) {\n            int xR = wR + yR * "+t+" - "+r+";\n\n            if (xR < 0 || xR >= "+e.inHeight+") {\n              continue;\n            }\n\n            for (int yC = 0; yC < "+e.outWidth+"; yC++) {\n              int xC = wC + yC * "+n+" - "+i+";\n\n              if (xC < 0 || xC >= "+e.inWidth+") {\n                continue;\n              }\n\n              float dyValue = getDy(b, yR, yC, d2);\n              float xValue = getX(b, xR, xC, d1);\n              dotProd += (xValue * dyValue);\n            }\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.Conv2DDerFilterProgram=r;var i=function(){return function(e){this.variableNames=["dy","W"],this.outputShape=e.inShape;var t=e.filterHeight,n=e.filterWidth,r=e.strideHeight,i=e.strideWidth,a=t-1-e.padInfo.top,o=n-1-e.padInfo.left;this.userCode="\n      const ivec2 pads = ivec2("+a+", "+o+");\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int batch = coords[0];\n        int d1 = coords[3];\n\n        ivec2 dyCorner = coords.yz - pads;\n        int dyRCorner = dyCorner.x;\n        int dyCCorner = dyCorner.y;\n\n        // Convolve dy(?, ?, d2) with w(:, :, d1, d2) to compute dx(xR, xC, d1).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n        for (int wR = 0; wR < "+t+"; wR++) {\n          float dyR = float(dyRCorner + wR) / "+r+".0;\n\n          if (dyR < 0.0 || dyR >= "+e.outHeight+".0 || fract(dyR) > 0.0) {\n            continue;\n          }\n          int idyR = int(dyR);\n\n          int wRPerm = "+t+" - 1 - wR;\n\n          for (int wC = 0; wC < "+n+"; wC++) {\n            float dyC = float(dyCCorner + wC) / "+i+".0;\n\n            if (dyC < 0.0 || dyC >= "+e.outWidth+".0 ||\n                fract(dyC) > 0.0) {\n              continue;\n            }\n            int idyC = int(dyC);\n\n            int wCPerm = "+n+" - 1 - wC;\n\n            for (int d2 = 0; d2 < "+e.outChannels+"; d2++) {\n              float xValue = getDy(batch, idyR, idyC, d2);\n              float wValue = getW(wRPerm, wCPerm, d1, d2);\n              dotProd += xValue * wValue;\n            }\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.Conv2DDerInputProgram=i},{}],19:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["x","W"],this.outputShape=e.outShape;var t=e.padInfo.top,n=e.padInfo.left,r=e.strideHeight,i=e.strideWidth,a=e.dilationHeight,o=e.dilationWidth,s=e.filterHeight,u=e.filterWidth,l=4*Math.floor(e.inChannels/4),c=e.inChannels%4;this.userCode="\n      const ivec2 strides = ivec2("+r+", "+i+");\n      const ivec2 pads = ivec2("+t+", "+n+");\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int batch = coords[0];\n        int d2 = coords[3];\n\n        ivec2 xRCCorner = coords.yz * strides - pads;\n        int xRCorner = xRCCorner.x;\n        int xCCorner = xRCCorner.y;\n\n        // Convolve x(?, ?, d1) with w(:, :, d1, d2) to get y(yR, yC, d2).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n        for (int wR = 0; wR < "+s+"; wR++) {\n          int xR = xRCorner + wR * "+a+";\n\n          if (xR < 0 || xR >= "+e.inHeight+") {\n            continue;\n          }\n\n          for (int wC = 0; wC < "+u+"; wC++) {\n            int xC = xCCorner + wC * "+o+";\n\n            if (xC < 0 || xC >= "+e.inWidth+") {\n              continue;\n            }\n\n            for (int d1 = 0; d1 < "+l+"; d1 += 4) {\n              vec4 xValues = vec4(\n                getX(batch, xR, xC, d1),\n                getX(batch, xR, xC, d1 + 1),\n                getX(batch, xR, xC, d1 + 2),\n                getX(batch, xR, xC, d1 + 3)\n              );\n              vec4 wValues = vec4(\n                getW(wR, wC, d1, d2),\n                getW(wR, wC, d1 + 1, d2),\n                getW(wR, wC, d1 + 2, d2),\n                getW(wR, wC, d1 + 3, d2)\n              );\n\n              dotProd += dot(xValues, wValues);\n            }\n\n            if ("+(1===c)+") {\n              dotProd +=\n                getX(batch, xR, xC, "+l+") *\n                getW(wR, wC, "+l+", d2);\n            } else if ("+(2===c)+") {\n              vec2 xValues = vec2(\n                getX(batch, xR, xC, "+l+"),\n                getX(batch, xR, xC, "+l+" + 1)\n              );\n              vec2 wValues = vec2(\n                getW(wR, wC, "+l+", d2),\n                getW(wR, wC, "+l+" + 1, d2)\n              );\n              dotProd += dot(xValues, wValues);\n            } else if ("+(3===c)+") {\n              vec3 xValues = vec3(\n                getX(batch, xR, xC, "+l+"),\n                getX(batch, xR, xC, "+l+" + 1),\n                getX(batch, xR, xC, "+l+" + 2)\n              );\n              vec3 wValues = vec3(\n                getW(wR, wC, "+l+", d2),\n                getW(wR, wC, "+l+" + 1, d2),\n                getW(wR, wC, "+l+" + 2, d2)\n              );\n              dotProd += dot(xValues, wValues);\n            }\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.Conv2DProgram=r},{}],20:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["x","W"],this.outputShape=e.outShape;var t=e.inHeight,n=e.inWidth,r=e.padInfo.top,i=e.padInfo.left,a=e.strideHeight,o=e.strideWidth,s=e.dilationHeight,u=e.dilationWidth,l=e.filterHeight,c=e.filterWidth,p=e.outChannels/e.inChannels;this.userCode="\n      const ivec2 strides = ivec2("+a+", "+o+");\n      const ivec2 pads = ivec2("+r+", "+i+");\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int batch = coords.x;\n        ivec2 xRCCorner = coords.yz * strides - pads;\n        int d2 = coords.w;\n        int d1 = d2 / "+p+";\n        int q = d2 - d1 * "+p+";\n\n        int xRCorner = xRCCorner.x;\n        int xCCorner = xRCCorner.y;\n\n        // Convolve x(?, ?, d1) with w(:, :, d1, q) to get y(yR, yC, d2).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n        // TODO(dsmilkov): Flatten the two for loops and vec4 the operations.\n        for (int wR = 0; wR < "+l+"; wR++) {\n          int xR = xRCorner + wR * "+s+";\n\n          if (xR < 0 || xR >= "+t+") {\n            continue;\n          }\n\n          for (int wC = 0; wC < "+c+"; wC++) {\n            int xC = xCCorner + wC * "+u+";\n\n            if (xC < 0 || xC >= "+n+") {\n              continue;\n            }\n\n            float xVal = getX(batch, xR, xC, d1);\n            float wVal = getW(wR, wC, d1, q);\n            dotProd += xVal * wVal;\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.DepthwiseConv2DProgram=r},{}],21:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["A"];var t=e[0],n=e[1];this.outputShape=e,this.userCode="\n      void main() {\n        ivec3 coords = getOutputCoords();\n        int texR = coords[0];\n        int texC = coords[1];\n        int depth = coords[2];\n        vec2 uv = (vec2(texC, texR) + halfCR) / vec2("+n+".0, "+t+".0);\n\n        vec4 values = texture2D(A, uv);\n        float value;\n        if (depth == 0) {\n          value = values.r;\n        } else if (depth == 1) {\n          value = values.g;\n        } else if (depth == 2) {\n          value = values.b;\n        } else if (depth == 3) {\n          value = values.a;\n        }\n\n        setOutput(floor(value * 255.0 + 0.5));\n      }\n    "}}();n.FromPixelsProgram=r},{}],22:[function(e,t,n){"use strict";function r(e,t){var n=e.length;if(n>4)throw Error("Gather for rank "+n+" is not yet supported");if(1===n)return"int(getIndices(resRC))";for(var r=["resRC.x","resRC.y","resRC.z","resRC.w"],i=[],a=0;a<e.length;a++)a===t?i.push("int(getIndices("+r[a]+"))"):i.push(""+r[a]);return i.join()}Object.defineProperty(n,"__esModule",{value:!0});var i=e("./shader_compiler"),a=function(){return function(e,t,n){this.variableNames=["A","indices"];var a=e.slice();a[n]=t,this.outputShape=a,this.rank=a.length;var o=i.getCoordsDataType(this.rank),s=r(e,n);this.userCode="\n      void main() {\n        "+o+" resRC = getOutputCoords();\n        setOutput(getA("+s+"));\n      }\n    "}}();n.GatherProgram=a},{"./shader_compiler":38}],23:[function(e,t,n){"use strict";var r=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},i=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var a=e("../../environment"),o=e("../../util"),s=e("./gpgpu_util"),u=e("./tex_util"),l=e("./webgl_util"),c=function(){function e(e){this.outputTexture=null,this.program=null,this.disposed=!1,this.autoDebugValidate=!1,this.vertexAttrsAreBound=!1,this.gl=null!=e?e:s.createWebGLContext(),1===a.ENV.get("WEBGL_VERSION")?(this.textureFloatExtension=l.getExtensionOrThrow(this.gl,"OES_texture_float"),this.colorBufferFloatExtension=this.gl.getExtension("WEBGL_color_buffer_float")):this.colorBufferFloatExtension=l.getExtensionOrThrow(this.gl,"EXT_color_buffer_float"),this.loseContextExtension=l.getExtensionOrThrow(this.gl,"WEBGL_lose_context"),a.ENV.get("WEBGL_GET_BUFFER_SUB_DATA_ASYNC_EXTENSION_ENABLED")&&(this.getBufferSubDataAsyncExtension=this.gl.getExtension("WEBGL_get_buffer_sub_data_async")),this.vertexBuffer=s.createVertexBuffer(this.gl),this.indexBuffer=s.createIndexBuffer(this.gl),this.framebuffer=l.createFramebuffer(this.gl)}return e.prototype.dispose=function(){var e=this;if(!this.disposed){null!=this.program&&console.warn("Disposing a GPGPUContext that still has a bound WebGLProgram. This is probably a resource leak, delete the program with GPGPUContext.deleteProgram before disposing."),null!=this.outputTexture&&console.warn("Disposing a GPGPUContext that still has a bound output matrix texture.  This is probably a resource leak, delete the output matrix texture with GPGPUContext.deleteMatrixTexture before disposing.");var t=this.gl;l.callAndCheck(t,function(){return t.finish()}),l.callAndCheck(t,function(){return t.bindFramebuffer(t.FRAMEBUFFER,null)}),l.callAndCheck(t,function(){return t.deleteFramebuffer(e.framebuffer)}),l.callAndCheck(t,function(){return t.bindBuffer(t.ARRAY_BUFFER,null)}),l.callAndCheck(t,function(){return t.deleteBuffer(e.vertexBuffer)}),l.callAndCheck(t,function(){return t.bindBuffer(t.ELEMENT_ARRAY_BUFFER,null)}),l.callAndCheck(t,function(){return t.deleteBuffer(e.indexBuffer)}),this.loseContextExtension.loseContext(),this.disposed=!0}},e.prototype.enableAutomaticDebugValidation=function(e){this.autoDebugValidate=e,l.enableDebugWebGLErrorChecking(e)},e.prototype.createMatrixTexture=function(e,t){return this.throwIfDisposed(),s.createMatrixTexture(this.gl,e,t)},e.prototype.uploadPixelDataToTexture=function(e,t){this.throwIfDisposed(),s.uploadPixelDataToTexture(this.gl,e,t)},e.prototype.createPackedMatrixTexture=function(e,t){return this.throwIfDisposed(),s.createPackedMatrixTexture(this.gl,e,t)},e.prototype.deleteMatrixTexture=function(e){var t=this;this.throwIfDisposed(),this.outputTexture===e&&(l.unbindColorTextureFromFramebuffer(this.gl,this.framebuffer),this.outputTexture=null),l.callAndCheck(this.gl,function(){return t.gl.deleteTexture(e)})},e.prototype.uploadMatrixToTexture=function(e,t,n,r){this.throwIfDisposed();return s.uploadMatrixToTexture(this.gl,e,t,n,r,1)},e.prototype.uploadMatrixToPackedTexture=function(e,t,n,r){return this.throwIfDisposed(),s.uploadMatrixToPackedTexture(this.gl,e,t,n,r)},e.prototype.downloadMatrixFromTexture=function(e,t,n){var r=this;return this.downloadMatrixDriver(e,function(){return s.downloadMatrixFromOutputTexture(r.gl,t,n)})},e.prototype.downloadMatrixFromTextureAsync=function(e,t,n){return r(this,void 0,void 0,function(){var r=this;return i(this,function(i){if(null==this.getBufferSubDataAsyncExtension)throw new Error("Cannot download matrix from output texture asynchronously, WEBGL_get_buffer_sub_data_async is not enabled.");return[2,this.downloadMatrixDriverAsync(e,function(){return s.downloadMatrixFromOutputTextureAsync(r.gl,r.getBufferSubDataAsyncExtension,t,n)})]})})},e.prototype.downloadMatrixFromRGBAColorTexture=function(e,t,n,r){var i=this;return this.downloadMatrixDriver(e,function(){return s.downloadMatrixFromRGBAColorTexture(i.gl,t,n,r)})},e.prototype.downloadMatrixFromPackedTexture=function(e,t,n){var r=this;return this.downloadMatrixDriver(e,function(){return s.downloadMatrixFromPackedOutputTexture(r.gl,t,n)})},e.prototype.createProgram=function(e){this.throwIfDisposed();var t=this.gl,n=l.createFragmentShader(t,e),r=s.createVertexShader(t),i=l.createProgram(t);return l.callAndCheck(t,function(){return t.attachShader(i,r)}),l.callAndCheck(t,function(){return t.attachShader(i,n)}),l.linkProgram(t,i),this.autoDebugValidate&&l.validateProgram(t,i),this.vertexAttrsAreBound||(this.setProgram(i),this.vertexAttrsAreBound=s.bindVertexProgramAttributeStreams(t,this.program,this.vertexBuffer)),i},e.prototype.deleteProgram=function(e){var t=this;this.throwIfDisposed(),e===this.program&&(this.program=null),null!=e&&l.callAndCheck(this.gl,function(){return t.gl.deleteProgram(e)})},e.prototype.setProgram=function(e){var t=this;this.throwIfDisposed(),this.program=e,null!=this.program&&this.autoDebugValidate&&l.validateProgram(this.gl,this.program),l.callAndCheck(this.gl,function(){return t.gl.useProgram(e)})},e.prototype.getUniformLocation=function(e,t,n){return void 0===n&&(n=!0),this.throwIfDisposed(),n?l.getProgramUniformLocationOrThrow(this.gl,e,t):l.getProgramUniformLocation(this.gl,e,t)},e.prototype.getAttributeLocation=function(e,t){var n=this;return this.throwIfDisposed(),l.callAndCheck(this.gl,function(){return n.gl.getAttribLocation(e,t)})},e.prototype.getUniformLocationNoThrow=function(e,t){return this.throwIfDisposed(),this.gl.getUniformLocation(e,t)},e.prototype.setInputMatrixTexture=function(e,t,n){this.throwIfDisposed(),this.throwIfNoProgram(),l.bindTextureToProgramUniformSampler(this.gl,this.program,e,t,n)},e.prototype.setOutputMatrixTexture=function(e,t,n){this.setOutputMatrixTextureDriver(e,n,t)},e.prototype.setOutputPackedMatrixTexture=function(e,t,n){this.throwIfDisposed();var r=u.getPackedMatrixTextureShapeWidthHeight(t,n),i=r[0],a=r[1];this.setOutputMatrixTextureDriver(e,i,a)},e.prototype.setOutputMatrixWriteRegion=function(e,t,n,r){this.setOutputMatrixWriteRegionDriver(n,e,r,t)},e.prototype.setOutputPackedMatrixWriteRegion=function(e,t,n,r){throw new Error("setOutputPackedMatrixWriteRegion not implemented.")},e.prototype.debugValidate=function(){null!=this.program&&l.validateProgram(this.gl,this.program),l.validateFramebuffer(this.gl)},e.prototype.executeProgram=function(){this.throwIfDisposed(),this.throwIfNoProgram();var e=this.gl;this.autoDebugValidate&&this.debugValidate(),l.callAndCheck(e,function(){return e.drawElements(e.TRIANGLES,6,e.UNSIGNED_SHORT,0)})},e.prototype.blockUntilAllProgramsCompleted=function(){var e=this;this.throwIfDisposed(),l.callAndCheck(this.gl,function(){return e.gl.finish()})},e.prototype.getQueryTimerExtension=function(){return null==this.disjointQueryTimerExtension&&(this.disjointQueryTimerExtension=l.getExtensionOrThrow(this.gl,2===a.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")?"EXT_disjoint_timer_query_webgl2":"EXT_disjoint_timer_query")),this.disjointQueryTimerExtension},e.prototype.getQueryTimerExtensionWebGL2=function(){return this.getQueryTimerExtension()},e.prototype.getQueryTimerExtensionWebGL1=function(){return this.getQueryTimerExtension()},e.prototype.runQuery=function(e){var t=this.beginQuery();return e(),this.endQuery(),this.pollQueryTime(t)},e.prototype.beginQuery=function(){if(2===a.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")){var e=this.gl,t=this.getQueryTimerExtensionWebGL2(),n=e.createQuery();return e.beginQuery(t.TIME_ELAPSED_EXT,n),n}var r=this.getQueryTimerExtensionWebGL1(),i=r.createQueryEXT();return r.beginQueryEXT(r.TIME_ELAPSED_EXT,i),i},e.prototype.endQuery=function(){if(2!==a.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION")){var e=this.getQueryTimerExtensionWebGL1();e.endQueryEXT(e.TIME_ELAPSED_EXT)}else{var t=this.gl,n=this.getQueryTimerExtensionWebGL2();t.endQuery(n.TIME_ELAPSED_EXT)}},e.prototype.isQueryAvailable=function(e,t){if(0===t)return!0;if(2===t){var n=this.gl,r=this.getQueryTimerExtensionWebGL2(),i=n.getQueryParameter(e,n.QUERY_RESULT_AVAILABLE),a=this.gl.getParameter(r.GPU_DISJOINT_EXT);return i&&!a}var i=(r=this.getQueryTimerExtensionWebGL1()).getQueryObjectEXT(e,r.QUERY_RESULT_AVAILABLE_EXT),a=this.gl.getParameter(r.GPU_DISJOINT_EXT);return i&&!a},e.prototype.pollQueryTime=function(e){var t=this;return new Promise(function(n,r){var i=a.ENV.get("WEBGL_DISJOINT_QUERY_TIMER_EXTENSION_VERSION");o.repeatedTry(function(){return t.isQueryAvailable(e,i)}).then(function(){return n(t.getQueryTime(e,i))}).catch(function(){console.warn("Disjoint query timer never available."),n(-1)})})},e.prototype.getQueryTime=function(e,t){if(0===t)return null;if(2===t){var n=this.gl;return(i=n.getQueryParameter(e,n.QUERY_RESULT))/1e6}var r=this.getQueryTimerExtensionWebGL1(),i=r.getQueryObjectEXT(e,r.QUERY_RESULT_EXT);return i/1e6},e.prototype.downloadMatrixDriverSetup=function(e){this.throwIfDisposed(),l.bindColorTextureToFramebuffer(this.gl,e,this.framebuffer),this.autoDebugValidate&&l.validateFramebuffer(this.gl)},e.prototype.downloadMatrixDriverTeardown=function(){null!=this.outputTexture?(l.bindColorTextureToFramebuffer(this.gl,this.outputTexture,this.framebuffer),this.autoDebugValidate&&l.validateFramebuffer(this.gl)):l.unbindColorTextureFromFramebuffer(this.gl,this.framebuffer)},e.prototype.downloadMatrixDriver=function(e,t){this.downloadMatrixDriverSetup(e);var n=t();return this.downloadMatrixDriverTeardown(),n},e.prototype.downloadMatrixDriverAsync=function(e,t){return r(this,void 0,void 0,function(){var n;return i(this,function(r){switch(r.label){case 0:return this.downloadMatrixDriverSetup(e),[4,t()];case 1:return n=r.sent(),this.downloadMatrixDriverTeardown(),[2,n]}})})},e.prototype.setOutputMatrixTextureDriver=function(e,t,n){this.throwIfDisposed();var r=this.gl;l.bindColorTextureToFramebuffer(r,e,this.framebuffer),this.autoDebugValidate&&l.validateFramebuffer(r),this.outputTexture=e,l.callAndCheck(r,function(){return r.viewport(0,0,t,n)}),l.callAndCheck(r,function(){return r.scissor(0,0,t,n)})},e.prototype.setOutputMatrixWriteRegionDriver=function(e,t,n,r){var i=this;this.throwIfDisposed(),l.callAndCheck(this.gl,function(){return i.gl.scissor(e,t,n,r)})},e.prototype.throwIfDisposed=function(){if(this.disposed)throw new Error("Attempted to use disposed GPGPUContext.")},e.prototype.throwIfNoProgram=function(){if(null==this.program)throw new Error("No GPU program is currently set.")},e}();n.GPGPUContext=c},{"../../environment":5,"../../util":95,"./gpgpu_util":25,"./tex_util":40,"./webgl_util":45}],24:[function(e,t,n){"use strict";function r(){return!a.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")}function i(e,t){if(e.length!==t.length)throw Error("Binary was compiled with "+e.length+" inputs, but was executed with "+t.length+" inputs");e.forEach(function(e,n){var r=e.logicalShape,i=e.texShape,a=t[n].tensor.shape,s=t[n].texData.texShape;if(!o.arraysEqual(r,a))throw Error("Binary was compiled with different shapes than the current args. Shapes "+r+" and "+a+" must match");if(!o.arraysEqual(i,s))throw Error("Binary was compiled with different texture shapes than the current args. Shape "+i+" and "+s+" must match")})}Object.defineProperty(n,"__esModule",{value:!0});var a=e("../../environment"),o=e("../../util"),s=e("./shader_compiler"),u="NaN";n.compileProgram=function(e,t,n,i){for(var a=t.userCode,o=n.map(function(e,n){var r={logicalShape:e.tensor.shape,texShape:e.texData.texShape};return{name:t.variableNames[n],shapeInfo:r}}),l=o.map(function(e){return e.shapeInfo}),c={logicalShape:i.tensor.shape,texShape:i.texData.texShape},p=s.makeShader(o,c,a,!0===t.supportsBroadcasting),h=e.createProgram(p),d={},f=0;f<t.variableNames.length;f++){var g=t.variableNames[f];d[g]=e.getUniformLocation(h,g)}r()&&(d[u]=e.getUniformLocation(h,u,!1));return{program:t,source:p,webGLProgram:h,uniformLocations:d,gpgpu:e,inShapeInfos:l,outShapeInfo:c}},n.runProgram=function(e,t,n,a){i(e.inShapeInfos,t),i([e.outShapeInfo],[n]);var o=n.texData.texture,s=n.texData.texShape,l=e.gpgpu;l.setOutputMatrixTexture(o,s[0],s[1]),l.setProgram(e.webGLProgram),t.forEach(function(t,n){var r=t.texData.texture,i=e.program.variableNames[n],a=e.uniformLocations[i];l.setInputMatrixTexture(r,a,n)}),r()&&l.gl.uniform1f(e.uniformLocations[u],NaN),null!=a&&a(l,e.webGLProgram),l.executeProgram()},n.makeShaderKey=function(e,t,n){var r="";t.concat(n).forEach(function(e){r+=e.tensor.shape+"_"+e.texData.texShape});var i=e.userCode,a=(!0===e.supportsBroadcasting).toString(),o=e.constructor.name;return o+="_"+a+"_"+r+"_"+i}},{"../../environment":5,"../../util":95,"./shader_compiler":38}],25:[function(e,t,n){"use strict";function r(){return{alpha:!1,antialias:!1,premultipliedAlpha:!1,preserveDrawingBuffer:!1,depth:!1,stencil:!1,failIfMajorPerformanceCaveat:!0}}function i(e,t){return d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")&&2===d.ENV.get("WEBGL_VERSION")?4===t?e.RGBA32F:e.R32F:e.RGBA}function a(e,t){return d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")&&2===d.ENV.get("WEBGL_VERSION")?4===t?e.RGBA:e.RED:e.RGBA}function o(e){return d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")?e.FLOAT:e.UNSIGNED_BYTE}function s(e,t,n,r){g.validateTextureSize(e,t,n);var s=g.createTexture(e),u=e.TEXTURE_2D,l=i(e,r),c=a(e,r);return g.callAndCheck(e,function(){return e.bindTexture(u,s)}),g.callAndCheck(e,function(){return e.texParameteri(u,e.TEXTURE_WRAP_S,e.CLAMP_TO_EDGE)}),g.callAndCheck(e,function(){return e.texParameteri(u,e.TEXTURE_WRAP_T,e.CLAMP_TO_EDGE)}),g.callAndCheck(e,function(){return e.texParameteri(u,e.TEXTURE_MIN_FILTER,e.NEAREST)}),g.callAndCheck(e,function(){return e.texParameteri(u,e.TEXTURE_MAG_FILTER,e.NEAREST)}),g.callAndCheck(e,function(){return e.texImage2D(u,0,l,t,n,0,c,o(e),null)}),g.callAndCheck(e,function(){return e.bindTexture(e.TEXTURE_2D,null)}),s}function u(e,t,n,r,i,s){var u=a(e,s);g.validateTextureSize(e,n,r),g.callAndCheck(e,function(){return e.bindTexture(e.TEXTURE_2D,t)}),g.callAndCheck(e,function(){return e.texSubImage2D(e.TEXTURE_2D,0,0,0,n,r,u,o(e),i)}),g.callAndCheck(e,function(){return e.bindTexture(e.TEXTURE_2D,null)})}function l(e,t,n){return d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")?new Float32Array(f.getUnpackedArraySizeFromMatrixSize(e*t,n)):new Uint8Array(e*t*n)}function c(e,t,n,r){if(d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")){var i=new Float32Array(t*n);return f.decodeMatrixFromUnpackedArray(e,i,r),i}return f.decodeToFloatArray(e)}var p=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},h=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var d=e("../../environment"),f=e("./tex_util"),g=e("./webgl_util");n.getWebGLContextAttributes=r,n.createWebGLContext=function(e){var t,n=r();return t=null!=e?g.createWebGLRenderingContextFromCanvas(e,n):g.createWebGLRenderingContext(n),g.callAndCheck(t,function(){return t.disable(t.DEPTH_TEST)}),g.callAndCheck(t,function(){return t.disable(t.STENCIL_TEST)}),g.callAndCheck(t,function(){return t.disable(t.BLEND)}),g.callAndCheck(t,function(){return t.disable(t.DITHER)}),g.callAndCheck(t,function(){return t.disable(t.POLYGON_OFFSET_FILL)}),g.callAndCheck(t,function(){return t.disable(t.SAMPLE_COVERAGE)}),g.callAndCheck(t,function(){return t.enable(t.SCISSOR_TEST)}),g.callAndCheck(t,function(){return t.enable(t.CULL_FACE)}),g.callAndCheck(t,function(){return t.cullFace(t.BACK)}),t},n.createVertexShader=function(e){return g.createVertexShader(e,"\n    precision highp float;\n    attribute vec3 clipSpacePos;\n    attribute vec2 uv;\n    varying vec2 resultUV;\n\n    void main() {\n      gl_Position = vec4(clipSpacePos, 1);\n      resultUV = uv;\n    }")},n.createVertexBuffer=function(e){var t=new Float32Array([-1,1,0,0,1,-1,-1,0,0,0,1,1,0,1,1,1,-1,0,1,0]);return g.createStaticVertexBuffer(e,t)},n.createIndexBuffer=function(e){var t=new Uint16Array([0,1,2,2,1,3]);return g.createStaticIndexBuffer(e,t)},n.createMatrixTexture=function(e,t,n){var r=f.getUnpackedMatrixTextureShapeWidthHeight(t,n);return s(e,r[0],r[1],1)},n.createColorMatrixTexture=function(e,t,n){var r=f.getColorMatrixTextureShapeWidthHeight(t,n);return s(e,r[0],r[1],4)},n.createPackedMatrixTexture=function(e,t,n){var r=f.getPackedMatrixTextureShapeWidthHeight(t,n);return s(e,r[0],r[1],4)},n.bindVertexProgramAttributeStreams=function(e,t,n){return g.callAndCheck(e,function(){return e.bindBuffer(e.ARRAY_BUFFER,n)}),g.bindVertexBufferToProgramAttribute(e,t,"clipSpacePos",n,3,20,0)&&g.bindVertexBufferToProgramAttribute(e,t,"uv",n,2,20,12)},n.uploadPixelDataToTexture=function(e,t,n){g.callAndCheck(e,function(){return e.bindTexture(e.TEXTURE_2D,t)}),g.callAndCheck(e,function(){return e.texImage2D(e.TEXTURE_2D,0,e.RGBA,e.RGBA,e.UNSIGNED_BYTE,n)}),g.callAndCheck(e,function(){return e.bindTexture(e.TEXTURE_2D,null)})},n.uploadMatrixToTexture=function(e,t,n,r,i,a){var o,s=f.getUnpackedMatrixTextureShapeWidthHeight(n,r),l=s[0],c=s[1];if(d.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")){var p=1===a?g.getChannelsPerTexture():a;1===p?o=i:(o=new Float32Array(f.getUnpackedArraySizeFromMatrixSize(i.length,p)),f.encodeMatrixToUnpackedArray(i,o,p))}else o=f.encodeFloatArray(i);u(e,t,l,c,o,a)},n.uploadMatrixToPackedTexture=function(e,t,n,r,i){var a=f.getPackedMatrixTextureShapeWidthHeight(n,r),o=a[0],s=a[1],l=new Float32Array(f.getPackedRGBAArraySizeFromMatrixShape(n,r));f.encodeMatrixToPackedRGBA(i,n,r,l);u(e,t,o,s,l,4)},n.downloadMatrixFromOutputTextureAsync=function(e,t,n,r){return p(this,void 0,void 0,function(){var i,a,s,u,p;return h(this,function(h){switch(h.label){case 0:return i=e,a=4,s=l(n,r,a),u=s instanceof Float32Array?4*s.length:s,p=e.createBuffer(),g.callAndCheck(e,function(){return e.bindBuffer(i.PIXEL_PACK_BUFFER,p)}),g.callAndCheck(e,function(){return e.bufferData(i.PIXEL_PACK_BUFFER,u,e.STATIC_DRAW)}),g.callAndCheck(e,function(){return i.readPixels(0,0,r,n,e.RGBA,o(e),0)}),[4,t.getBufferSubDataAsync(i.PIXEL_PACK_BUFFER,0,s)];case 1:return h.sent(),[2,c(s,n,r,a)]}})})},n.downloadMatrixFromOutputTexture=function(e,t,n){var r=f.getUnpackedMatrixTextureShapeWidthHeight(t,n),i=r[0],a=r[1],s=l(t,n,4);return g.callAndCheck(e,function(){return e.readPixels(0,0,i,a,e.RGBA,o(e),s)}),c(s,t,n,4)},n.downloadMatrixFromRGBAColorTexture=function(e,t,n,r){var i=t*n*4,a=new Uint8Array(i);g.callAndCheck(e,function(){return e.readPixels(0,0,n,t,e.RGBA,e.UNSIGNED_BYTE,a)});for(var o=new Float32Array(i),s=0;s<a.length;s++)o[s]=a[s];var u=new Float32Array(t*n*r);return f.decodeMatrixFromUnpackedColorRGBAArray(o,u,r),u},n.downloadMatrixFromPackedOutputTexture=function(e,t,n){var r=f.getPackedMatrixTextureShapeWidthHeight(t,n),i=r[0],a=r[1],s=new Float32Array(f.getPackedRGBAArraySizeFromMatrixShape(t,n));g.callAndCheck(e,function(){return e.readPixels(0,0,i,a,e.RGBA,o(e),s)});var u=new Float32Array(t*n);return f.decodeMatrixFromPackedRGBA(s,t,n,u)}},{"../../environment":5,"./tex_util":40,"./webgl_util":45}],26:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./shader_compiler"),i=function(){return function(e,t,n){this.variableNames=["c","a","b"],this.outputShape=t;var i,a;if(n>4)throw Error("Where for rank "+n+" is not yet supported");if(1===n)a="resRC",i="resRC";else{for(var o=["resRC.x","resRC.y","resRC.z","resRC.w"],s=[],u=[],l=0;l<t.length;l++)u.push(""+o[l]),l<e&&s.push(""+o[l]);i=s.join(),a=u.join()}var c=r.getCoordsDataType(n);this.userCode="\n      void main() {\n        "+c+" resRC = getOutputCoords();\n        float cVal = getC("+i+");\n        if (cVal >= 1.0) {\n          setOutput(getA("+a+"));\n        } else {\n          setOutput(getB("+a+"));\n        }\n      }\n    "}}();n.WhereProgram=i},{"./shader_compiler":38}],27:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n,r,i){this.variableNames=["x"],this.outputShape=[];var a=t,o=e[3]-1;this.outputShape=e;var s,u="float("+n+") + float("+r+") * sum";s=.5===i?"inversesqrt("+u+")":1===i?"1.0/("+u+")":"exp(log("+u+") * float(-"+i+"));",this.userCode="\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int b = coords[0];\n        int r = coords[1];\n        int c = coords[2];\n        int d = coords[3];\n        float x = getX(b, r, c, d);\n        float sum = 0.0;\n        for (int j = -"+a+"; j <= "+a+"; j++) {\n          int idx = d + j;\n          if (idx >= 0 && idx <=  "+o+") {\n            float z = getX(b, r, c, idx);\n            sum += z * z;\n          }\n        }\n        float val = x * "+s+";\n        setOutput(val);\n      }\n    "}}();n.LRNProgram=r},{}],28:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e){this.variableNames=["dy","maxPos"],this.outputShape=e.inShape;var t=e.filterHeight,n=e.filterWidth,r=e.strideHeight,i=e.strideWidth,a=t-1-e.padInfo.top,o=n-1-e.padInfo.left,s=t*n-1;this.userCode="\n      const ivec2 pads = ivec2("+a+", "+o+");\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int b = coords[0];\n        int d = coords[3];\n\n        ivec2 dyRCCorner = coords.yz - pads;\n        int dyRCorner = dyRCCorner.x;\n        int dyCCorner = dyRCCorner.y;\n\n        // Convolve dy(?, ?, d) with pos mask(:, :, d) to get dx(xR, xC, d).\n        // ? = to be determined. : = across all values in that axis.\n        float dotProd = 0.0;\n        for (int wR = 0; wR < "+t+"; wR++) {\n          float dyR = float(dyRCorner + wR) / "+r+".0;\n\n          if (dyR < 0.0 || dyR >= "+e.outHeight+".0 || fract(dyR) > 0.0) {\n            continue;\n          }\n          int idyR = int(dyR);\n\n          for (int wC = 0; wC < "+n+"; wC++) {\n            float dyC = float(dyCCorner + wC) / "+i+".0;\n\n            if (dyC < 0.0 || dyC >= "+e.outWidth+".0 ||\n                fract(dyC) > 0.0) {\n              continue;\n            }\n            int idyC = int(dyC);\n\n            float dyValue = getDy(b, idyR, idyC, d);\n            int maxPosValue = "+s+" - int(getMaxPos(b, idyR, idyC, d));\n\n            // Get the current value, check it against the value from the\n            // position matrix.\n            int curPosValue = wR * "+n+" + wC;\n            float mask = float(maxPosValue == curPosValue ? 1.0 : 0.0);\n\n            dotProd += dyValue * mask;\n          }\n        }\n        setOutput(dotProd);\n      }\n    "}}();n.MaxPool2DBackpropProgram=r},{}],29:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n,r){void 0===n&&(n=!1),void 0===r&&(r=!1),this.variableNames=["matrixA","matrixB"];var i=n?e[1]:e[0],a=r?t[0]:t[1],o=n?e[0]:e[1];this.outputShape=[i,a];var s=function(e,t){return n?t+" + "+e+", aRow":"aRow, "+t+" + "+e},u=function(e,t){return r?"bCol, "+t+" + "+e:t+" + "+e+", bCol"},l=4*Math.floor(o/4),c=o%4;this.userCode=" float dotARowBCol(int aRow, int bCol) {\n      float result = 0.0;\n      for (int i = 0; i < "+l+"; i += 4) {\n        vec4 a = vec4(\n          getMatrixA("+s(0,"i")+"),\n          getMatrixA("+s(1,"i")+"),\n          getMatrixA("+s(2,"i")+"),\n          getMatrixA("+s(3,"i")+")\n        );\n        vec4 b = vec4(\n          getMatrixB("+u(0,"i")+"),\n          getMatrixB("+u(1,"i")+"),\n          getMatrixB("+u(2,"i")+"),\n          getMatrixB("+u(3,"i")+")\n        );\n\n        result += dot(a, b);\n      }\n\n      if ("+(1===c)+") {\n        result += getMatrixA("+s(0,l)+") *\n          getMatrixB("+u(0,l)+");\n      } else if ("+(2===c)+") {\n        vec2 a = vec2(\n          getMatrixA("+s(0,l)+"),\n          getMatrixA("+s(1,l)+")\n        );\n        vec2 b = vec2(\n          getMatrixB("+u(0,l)+"),\n          getMatrixB("+u(1,l)+")\n        );\n        result += dot(a, b);\n      } else if ("+(3===c)+") {\n        vec3 a = vec3(\n          getMatrixA("+s(0,l)+"),\n          getMatrixA("+s(1,l)+"),\n          getMatrixA("+s(2,l)+")\n        );\n        vec3 b = vec3(\n          getMatrixB("+u(0,l)+"),\n          getMatrixB("+u(1,l)+"),\n          getMatrixB("+u(2,l)+")\n        );\n        result += dot(a, b);\n      }\n\n      return result;\n    }\n\n    void main() {\n      ivec2 resRC = getOutputCoords();\n      setOutput(dotARowBCol(resRC.x, resRC.y));\n    }\n    "}}();n.MatMulProgram=r},{}],30:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){function e(e,t,n){this.variableNames=["probs"],this.outputShape=[e,n],this.userCode="\n      uniform float seed;\n\n      void main() {\n        ivec2 coords = getOutputCoords();\n        int batch = coords[0];\n\n        float r = random(seed);\n        float cdf = 0.0;\n\n        for (int i = 0; i < "+(t-1)+"; i++) {\n          cdf += getProbs(batch, i);\n\n          if (r < cdf) {\n            setOutput(float(i));\n            return;\n          }\n        }\n\n        // If no other event happened, last event happened.\n        setOutput(float("+(t-1)+"));\n      }\n    "}return e.prototype.getCustomSetupFunc=function(e){var t=this;return function(n,r){null==t.seedLoc&&(t.seedLoc=n.getUniformLocation(r,"seed")),n.gl.uniform1f(t.seedLoc,e)}},e}();n.MultinomialProgram=r},{}],31:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n,r){this.variableNames=["indices"],this.outputShape=[e,t],this.userCode="\n      void main() {\n        ivec2 coords = getOutputCoords();\n        int index = round(getIndices(coords.x));\n        setOutput(mix(float("+r+"), float("+n+"),\n                      float(index == coords.y)));\n      }\n    "}}();n.OneHotProgram=r},{}],32:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./shader_compiler"),i=function(){return function(e,t,n){this.variableNames=["x"],this.outputShape=t.map(function(t,n){return t[0]+e[n]+t[1]});var i=e.length,a=r.getCoordsDataType(i),o=t.map(function(e){return e[0]}).join(","),s=t.map(function(t,n){return t[0]+e[n]}).join(","),u=["coords[0]","coords[1]","coords[2]","coords[3]"].slice(0,i);this.userCode=1!==i?"\n      "+a+" start = "+a+"("+o+");\n      "+a+" end = "+a+"("+s+");\n\n      void main() {\n        "+a+" outC = getOutputCoords();\n        if (any(lessThan(outC, start)) || any(greaterThanEqual(outC, end))) {\n          setOutput(float("+n+"));\n        } else {\n          "+a+" coords = outC - start;\n          setOutput(getX("+u+"));\n        }\n      }\n    ":"\n        int start = "+o+";\n        int end = "+s+";\n\n        void main() {\n          int outC = getOutputCoords();\n          if (outC < start || outC >= end) {\n            setOutput(float("+n+"));\n          } else {\n            setOutput(getX(outC - start));\n          }\n        }\n      "}}();n.PadProgram=i},{"./shader_compiler":38}],33:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n){if(this.variableNames=["x"],"avg"===t&&n)throw new Error("Cannot compute positions for average pool.");var r=e.filterHeight,i=e.filterWidth,a=e.strideHeight,o=e.strideWidth,s=e.padInfo.top,u=e.padInfo.left;this.outputShape=e.outShape;var l="avg"===t,c="0.0";if(l||(c="-1.0 / 0.0"),n)this.userCode="\n        const ivec2 strides = ivec2("+a+", "+o+");\n        const ivec2 pads = ivec2("+s+", "+u+");\n\n        void main() {\n          ivec4 coords = getOutputCoords();\n          int batch = coords[0];\n          int d = coords[3];\n\n          ivec2 xRCCorner = coords.yz * strides - pads;\n          int xRCorner = xRCCorner.x;\n          int xCCorner = xRCCorner.y;\n\n          // max/min x(?, ?, d) to get y(yR, yC, d).\n          // ? = to be determined\n          float minMaxValue = 0.0;\n          float minMaxValueFound = 0.0;\n          int minMaxPosition = 0;\n          float avgValue = 0.0;\n\n          for (int wR = 0; wR < "+r+"; wR++) {\n            int xR = xRCorner + wR;\n\n            if (xR < 0 || xR >= "+e.inHeight+") {\n              continue;\n            }\n\n            for (int wC = 0; wC < "+i+"; wC++) {\n              int xC = xCCorner + wC;\n\n              if (xC < 0 || xC >= "+e.inWidth+") {\n                continue;\n              }\n\n              float value = getX(batch, xR, xC, d);\n\n              // If a min / max value has already been found, use it. If not,\n              // use the current value.\n              float currMinMaxValue = mix(\n                  value, minMaxValue, minMaxValueFound);\n              if (value >= currMinMaxValue) {\n                minMaxValue = value;\n                minMaxValueFound = 1.0;\n                minMaxPosition = wR * "+i+" + wC;\n              }\n            }\n          }\n          setOutput(float(minMaxPosition));\n        }\n      ";else{var p=t+"("+t+"("+t+"(minMaxValue[0], minMaxValue[1]), minMaxValue[2]), minMaxValue[3])";"avg"===t&&(p="avgValue / count");var h=4*Math.floor(i/4),d=i%4,f="\n      if ("+l+") {\n        avgValue += dot(values, ones);\n      } else {\n        minMaxValue = max(values, minMaxValue);\n      }\n    ";this.userCode="\n      const ivec2 strides = ivec2("+a+", "+o+");\n      const ivec2 pads = ivec2("+s+", "+u+");\n      const float initializationValue = "+c+";\n      const vec4 ones = vec4(1.0, 1.0, 1.0, 1.0);\n\n      float count = 0.0;\n\n      float getValue(int batch, int xR, int xC, int d) {\n        if (xC < 0 || xC >= "+e.inWidth+") {\n          return initializationValue;\n        }\n        count += 1.0;\n        return getX(batch, xR, xC, d);\n      }\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int batch = coords[0];\n        int d = coords[3];\n\n        ivec2 xRCCorner = coords.yz * strides - pads;\n        int xRCorner = xRCCorner.x;\n        int xCCorner = xRCCorner.y;\n\n        // max/min x(?, ?, d) to get y(yR, yC, d).\n        // ? = to be determined\n        vec4 minMaxValue = vec4("+c+");\n        float avgValue = 0.0;\n        count = 0.0;\n\n        for (int wR = 0; wR < "+r+"; wR++) {\n          int xR = xRCorner + wR;\n\n          if (xR < 0 || xR >= "+e.inHeight+") {\n            continue;\n          }\n\n          for (int wC = 0; wC < "+h+"; wC += 4) {\n            int xC = xCCorner + wC;\n\n            vec4 values = vec4(\n              getValue(batch, xR, xC, d),\n              getValue(batch, xR, xC + 1, d),\n              getValue(batch, xR, xC + 2, d),\n              getValue(batch, xR, xC + 3, d)\n            );\n\n            "+f+"\n          }\n\n          int xC = xCCorner + "+h+";\n          if ("+(1===d)+") {\n            vec4 values = vec4(\n              getValue(batch, xR, xC, d),\n              initializationValue,\n              initializationValue,\n              initializationValue\n            );\n\n            "+f+"\n          } else if ("+(2===d)+") {\n            vec4 values = vec4(\n              getValue(batch, xR, xC, d),\n              getValue(batch, xR, xC + 1, d),\n              initializationValue,\n              initializationValue\n            );\n\n            "+f+"\n          } else if ("+(3===d)+") {\n            vec4 values = vec4(\n              getValue(batch, xR, xC, d),\n              getValue(batch, xR, xC + 1, d),\n              getValue(batch, xR, xC + 2, d),\n              initializationValue\n            );\n\n            "+f+"\n          }\n        }\n        setOutput("+p+");\n      }\n    "}}}();n.Pool2DProgram=r},{}],34:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t){this.variableNames=["x"];var n=e.windowSize,r=e.batchSize,i=e.inSize,a=Math.ceil(i/n);this.outputShape=[r,a];var o="sum"===t,s="0.0";o||(s="min"===t?"1.0 / 0.0":"-1.0 / 0.0");var u="min"===t?"min":"max",l=t+"("+t+"("+t+"(minMaxValue[0], minMaxValue[1]), minMaxValue[2]), minMaxValue[3])";"sum"===t&&(l="sumValue");var c=4*Math.floor(n/4),p=n%4,h="\n      if ("+o+") {\n        sumValue += dot(values, ones);\n      } else {\n        minMaxValue = "+u+"(values, minMaxValue);\n      }\n    ",d="";i%n>0&&(d="\n        if (inIdx < 0 || inIdx >= "+i+") {\n          return initializationValue;\n        }\n      "),this.userCode="\n      const float initializationValue = "+s+";\n      const vec4 ones = vec4(1.0, 1.0, 1.0, 1.0);\n\n      float getValue(int batch, int inIdx) {\n        "+d+"\n        return getX(batch, inIdx);\n      }\n\n      void main() {\n        ivec2 coords = getOutputCoords();\n        int batch = coords[0];\n        int outIdx = coords[1];\n        int inOffset = outIdx * "+n+";\n\n        vec4 minMaxValue = vec4("+s+");\n        float sumValue = 0.0;\n\n        for (int i = 0; i < "+c+"; i += 4) {\n          int inIdx = inOffset + i;\n          vec4 values = vec4(\n            getValue(batch, inIdx),\n            getValue(batch, inIdx + 1),\n            getValue(batch, inIdx + 2),\n            getValue(batch, inIdx + 3)\n          );\n\n          "+h+"\n        }\n\n        int inIdx = inOffset + "+c+";\n        if ("+(1===p)+") {\n          vec4 values = vec4(\n            getValue(batch, inIdx),\n            initializationValue,\n            initializationValue,\n            initializationValue\n          );\n          "+h+"\n        } else if ("+(2===p)+") {\n          vec4 values = vec4(\n            getValue(batch, inIdx),\n            getValue(batch, inIdx + 1),\n            initializationValue,\n            initializationValue\n          );\n          "+h+"\n        } else if ("+(3===p)+") {\n          vec4 values = vec4(\n            getValue(batch, inIdx),\n            getValue(batch, inIdx + 1),\n            getValue(batch, inIdx + 2),\n            initializationValue\n          );\n          "+h+"\n        }\n        setOutput("+l+");\n      }\n    "}}();n.ReduceProgram=r},{}],35:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n,r){this.variableNames=["A"],this.outputShape=[];var i=e[0],a=e[1],o=e[2],s=e[3];this.outputShape=[i,t,n,s];var u=r?[a-1,o-1]:[a,o],l=r?[t-1,n-1]:[t,n];this.userCode="\n      const vec2 effectiveInputOverOutputRatioRC = vec2(\n          "+u[0]/l[0]+",\n          "+u[1]/l[1]+");\n      const vec2 inputShapeRC = vec2("+a+".0, "+o+".0);\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int b = coords[0];\n        int d = coords[3];\n        ivec2 yRC = coords.yz;\n\n        // Fractional source index.\n        vec2 sourceFracIndexRC = vec2(yRC) * effectiveInputOverOutputRatioRC;\n\n        // Compute the four integer indices.\n        ivec2 sourceFloorRC = ivec2(sourceFracIndexRC);\n        ivec2 sourceCeilRC = ivec2(\n          min(inputShapeRC - 1.0, ceil(sourceFracIndexRC)));\n\n        float topLeft = getA(b, sourceFloorRC.x, sourceFloorRC.y, d);\n        float bottomLeft = getA(b, sourceCeilRC.x, sourceFloorRC.y, d);\n        float topRight = getA(b, sourceFloorRC.x, sourceCeilRC.y, d);\n        float bottomRight = getA(b, sourceCeilRC.x, sourceCeilRC.y, d);\n\n        vec2 fracRC = sourceFracIndexRC - vec2(sourceFloorRC);\n\n        float top = topLeft + (topRight - topLeft) * fracRC.y;\n        float bottom = bottomLeft + (bottomRight - bottomLeft) * fracRC.y;\n        float newValue = top + (bottom - top) * fracRC.x;\n\n        setOutput(newValue);\n      }\n    "}}();n.ResizeBilinearProgram=r},{}],36:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=function(){return function(e,t,n,r){this.variableNames=["A"],this.outputShape=[];var i=e[0],a=e[1],o=e[2],s=e[3];this.outputShape=[i,t,n,s];var u=r?[a-1,o-1]:[a,o],l=r?[t-1,n-1]:[t,n];this.userCode="\n      const vec2 effectiveInputOverOutputRatioRC = vec2(\n          "+u[0]/l[0]+",\n          "+u[1]/l[1]+");\n      const vec2 inputShapeRC = vec2("+a+".0, "+o+".0);\n\n      void main() {\n        ivec4 coords = getOutputCoords();\n        int b = coords[0];\n        int d = coords[3];\n        ivec2 yRC = coords.yz;\n\n        // Fractional source index.\n        vec2 sourceFracIndexRC = vec2(yRC) * effectiveInputOverOutputRatioRC;\n\n        // Compute the coordinators of nearest neighbor point.\n        ivec2 sourceNearestRC = ivec2(\n          min(inputShapeRC - 1.0, floor(sourceFracIndexRC + 0.5)));\n\n        float newValue = getA(b, sourceNearestRC.x, sourceNearestRC.y, d);\n\n        setOutput(newValue);\n      }\n    "}}();n.ResizeNearestNeighborProgram=r},{}],37:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./shader_compiler"),i=function(){return function(e,t){this.variableNames=["x"];var n=e.length;if(n>4)throw new Error("WebGL backend: Reverse of rank-"+n+" tensor is not yet supported");if(this.outputShape=e,1!==n){var i=function(n){return-1!==t.indexOf(n)&&1!==e[n]?e[n]+" - coords["+n+"] - 1":"coords["+n+"]"},a=e.map(function(e,t){return i(t)}).join(","),o=r.getCoordsDataType(n);this.userCode="\n      void main() {\n        "+o+" coords = getOutputCoords();\n        setOutput(getX("+a+"));\n      }\n    "}else this.userCode="\n        void main() {\n          int coord = getOutputCoords();\n          setOutput(getX("+e[0]+" - coord - 1));\n        }\n      "}}();n.ReverseProgram=i},{"./shader_compiler":38}],38:[function(e,t,n){"use strict";function r(){return A.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")?O:S}function i(){return A.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")?k:C}function a(e){var t=e.shapeInfo.logicalShape;switch(t.length){case 0:return h(e);case 1:return d(e);case 2:return f(e);case 3:return g(e);case 4:return m(e);default:throw new Error(t.length+"-D input sampling is not yet supported")}}function o(e,t,n){var r=v(e);return r+=a(e),(n||E.arraysEqual(e.shapeInfo.logicalShape,t.logicalShape))&&(r+=b(e,t,n)),r}function s(e,t){switch(e.length){case 0:return"\n    int getOutputCoords() {\n      return 0;\n    }\n  ";case 1:return u(e,t);case 2:return p(e,t);case 3:return l(e,t);case 4:return c(e,t);default:throw new Error(e.length+"-D output sampling is not yet supported")}}function u(e,t){return 1===t[0]?"\n      int getOutputCoords() {\n        return int(resultUV.x * "+t[1]+".0);\n      }\n    ":1===t[1]?"\n      int getOutputCoords() {\n        return int(resultUV.y * "+t[0]+".0);\n      }\n    ":"\n    int getOutputCoords() {\n      ivec2 resTexRC = ivec2(resultUV.yx *\n                             vec2("+t[0]+", "+t[1]+"));\n      return resTexRC.x * "+t[1]+" + resTexRC.y;\n    }\n  "}function l(e,t){var n=e[1]*e[2],r=e[2];return"\n    ivec3 getOutputCoords() {\n      ivec2 resTexRC = ivec2(resultUV.yx *\n                             vec2("+t[0]+", "+t[1]+"));\n      int index = resTexRC.x * "+t[1]+" + resTexRC.y;\n      int r = index / "+n+";\n      index -= r * "+n+";\n      int c = index / "+r+";\n      int d = index - c * "+r+";\n      return ivec3(r, c, d);\n    }\n  "}function c(e,t){var n=e[3],r=e[2]*n,i=e[1]*r;return"\n    ivec4 getOutputCoords() {\n      ivec2 resTexRC = ivec2(resultUV.yx *\n        vec2("+t[0]+", "+t[1]+"));\n      int index = resTexRC.x * "+t[1]+" + resTexRC.y;\n\n      int r = index / "+i+";\n      index -= r * "+i+";\n\n      int c = index / "+r+";\n      index -= c * "+r+";\n\n      int d = index / "+n+";\n      int d2 = index - d * "+n+";\n\n      return ivec4(r, c, d, d2);\n    }\n  "}function p(e,t){return E.arraysEqual(e,t)?"\n      ivec2 getOutputCoords() {\n        return ivec2(resultUV.yx * vec2("+t[0]+", "+t[1]+"));\n      }\n    ":1===e[1]?"\n      ivec2 getOutputCoords() {\n        ivec2 resTexRC = ivec2(resultUV.yx *\n                               vec2("+t[0]+", "+t[1]+"));\n        int index = resTexRC.x * "+t[1]+" + resTexRC.y;\n        return ivec2(index, 0);\n      }\n    ":1===e[0]?"\n      ivec2 getOutputCoords() {\n        ivec2 resTexRC = ivec2(resultUV.yx *\n                               vec2("+t[0]+", "+t[1]+"));\n        int index = resTexRC.x * "+t[1]+" + resTexRC.y;\n        return ivec2(0, index);\n      }\n    ":"\n    ivec2 getOutputCoords() {\n      ivec2 resTexRC = ivec2(resultUV.yx *\n                             vec2("+t[0]+", "+t[1]+"));\n      int index = resTexRC.x * "+t[1]+" + resTexRC.y;\n      int r = index / "+e[1]+";\n      int c = index - r * "+e[1]+";\n      return ivec2(r, c);\n    }\n  "}function h(e){var t=e.name;return"\n    float "+("get"+t.charAt(0).toUpperCase()+t.slice(1))+"() {\n      return sampleTexture("+t+", halfCR);\n    }\n  "}function d(e){var t=e.name,n="get"+t.charAt(0).toUpperCase()+t.slice(1);return"\n    float "+n+"(int index) {\n      return "+n+"Flat(index);\n    }\n  "}function f(e){var t=e.shapeInfo.logicalShape,n=e.shapeInfo.texShape,r=e.name,i="get"+r.charAt(0).toUpperCase()+r.slice(1),o=n[0],s=n[1];if(E.arraysEqual(t,n))return"\n    float "+i+"(int row, int col) {\n      vec2 uv = (vec2(col, row) + halfCR) / vec2("+s+".0, "+o+".0);\n      return sampleTexture("+r+", uv);\n    }\n  ";var u=E.squeezeShape(t),l=u.newShape,c=u.keptDims,p=l;if(p.length<t.length){var h=["row","col"];return"\n      "+a(w(e,p))+"\n      float "+i+"(int row, int col) {\n        return "+i+"("+x(h,c)+");\n      }\n    "}return 1===s?"\n    float "+i+"(int row, int col) {\n      int index = row * "+t[1]+" + col;\n      vec2 uv = vec2(0.5, (float(index) + 0.5) / "+o+".0);\n      return sampleTexture("+r+", uv);\n    }\n  ":1===o?"\n    float "+i+"(int row, int col) {\n      int index = row * "+t[1]+" + col;\n      vec2 uv = vec2((float(index) + 0.5) / "+s+".0, 0.5);\n      return sampleTexture("+r+", uv);\n    }\n  ":"\n  float "+i+"(int row, int col) {\n    vec2 uv = UVfrom2D("+o+", "+s+", "+t[1]+", row, col);\n    return sampleTexture("+r+", uv);\n  }\n"}function g(e){var t=e.shapeInfo.texShape,n=e.shapeInfo.logicalShape,r=e.name,i="get"+r.charAt(0).toUpperCase()+r.slice(1),o=t[0],s=t[1],u=n[1]*n[2],l=n[2],c=E.squeezeShape(n),p=c.newShape,h=c.keptDims,d=p;if(d.length<n.length){var f=["row","col","depth"];return"\n        "+a(w(e,d))+"\n        float "+i+"(int row, int col, int depth) {\n          return "+i+"("+x(f,h)+");\n        }\n      "}return s===u?"\n        float "+i+"(int row, int col, int depth) {\n          int texR = row;\n          int texC = col * "+l+" + depth;\n          vec2 uv = (vec2(texC, texR) + halfCR) /\n                     vec2("+s+".0, "+o+".0);\n          return sampleTexture("+r+", uv);\n        }\n      ":s===l?"\n    float "+i+"(int row, int col, int depth) {\n      int texR = row * "+n[1]+" + col;\n      int texC = depth;\n      vec2 uv = (vec2(texC, texR) + halfCR) / vec2("+s+".0, "+o+".0);\n      return sampleTexture("+r+", uv);\n    }\n  ":"\n      float "+i+"(int row, int col, int depth) {\n        vec2 uv = UVfrom3D(\n            "+o+", "+s+", "+u+", "+l+", row, col, depth);\n        return sampleTexture("+r+", uv);\n      }\n  "}function m(e){var t=e.shapeInfo.logicalShape,n=e.shapeInfo.texShape,r=e.name,i="get"+r.charAt(0).toUpperCase()+r.slice(1),o=n[0],s=n[1],u=t[3],l=t[2]*u,c=t[1]*l,p=E.squeezeShape(t),h=p.newShape,d=p.keptDims;if(h.length<t.length){var f=["row","col","depth","depth2"];return"\n      "+a(w(e,h))+"\n      float "+i+"(int row, int col, int depth, int depth2) {\n        return "+i+"("+x(f,d)+");\n      }\n    "}return s===c?"\n      float "+i+"(int row, int col, int depth, int depth2) {\n        int texR = row;\n        int texC = col * "+l+" + depth * "+u+" + depth2;\n        vec2 uv = (vec2(texC, texR) + halfCR) /\n                   vec2("+s+".0, "+o+".0);\n        return sampleTexture("+r+", uv);\n      }\n    ":s===u?"\n      float "+i+"(int row, int col, int depth, int depth2) {\n        int texR = row * "+t[1]*t[2]+" + col * "+t[2]+" + depth;\n        int texC = depth2;\n        vec2 uv = (vec2(texC, texR) + halfCR) /\n                  vec2("+s+".0, "+o+".0);\n        return sampleTexture("+r+", uv);\n      }\n    ":"\n    float "+i+"(int row, int col, int depth, int depth2) {\n      vec2 uv = UVfrom4D("+o+", "+s+", "+c+", "+l+",\n          "+u+", row, col, depth, depth2);\n      return sampleTexture("+r+", uv);\n    }\n  "}function v(e){var t=e.name,n=e.shapeInfo.texShape,r="get"+t.charAt(0).toUpperCase()+t.slice(1)+"Flat",i=n[0],a=n[1];return 1===a&&1===i?"\n      float "+r+"(int index) {\n        return sampleTexture("+t+", halfCR);\n      }\n    ":1===a?"\n      float "+r+"(int index) {\n        vec2 uv = vec2(0.5, (float(index) + 0.5) / "+i+".0);\n        return sampleTexture("+t+", uv);\n      }\n    ":1===i?"\n      float "+r+"(int index) {\n        vec2 uv = vec2((float(index) + 0.5) / "+a+".0, 0.5);\n        return sampleTexture("+t+", uv);\n      }\n    ":"\n    float "+r+"(int index) {\n      vec2 uv = UVfrom1D("+i+", "+a+", index);\n      return sampleTexture("+t+", uv);\n    }\n  "}function y(e,t,n,r){var i=e.shapeInfo.logicalShape.length,a=t.logicalShape.length,o="int";2===a?o="ivec2":3===a?o="ivec3":4===a&&(o="ivec4");var s,u=_.getBroadcastDims(e.shapeInfo.logicalShape,t.logicalShape),l=a-i;s=0===i?"":a<2&&u.length>=1?"coords = 0;":u.map(function(e){return"coords["+(e+l)+"] = 0;"}).join("\n");var c="";return c=a<2&&i>0?"coords":e.shapeInfo.logicalShape.map(function(e,t){return"coords["+(t+l)+"]"}).join(", "),"\n    float "+r+"() {\n      "+o+" coords = getOutputCoords();\n      "+s+"\n      return get"+n+"("+c+");\n    }\n  "}function b(e,t,n){var r=e.shapeInfo.texShape,i=e.name,a=i.charAt(0).toUpperCase()+i.slice(1),o="get"+a+"AtOutCoords",s=_.getBroadcastDims(e.shapeInfo.logicalShape,t.logicalShape),u=e.shapeInfo.logicalShape.length,l=t.logicalShape.length,c=n&&(l>u||s.length>0),p=_.broadcastDimsAreOuter(s);if(c&&!p)return y(e,t,a,o);var h=t.texShape;if(E.arraysEqual(r,h))return"\n      float "+o+"() {\n        return sampleTexture("+i+", resultUV);\n      }\n    ";var d=E.sizeFromShape(r),f="";return c&&p&&(f="\n        int mainPart = index / "+d+";\n        index -= mainPart * "+d+";\n      "),"\n    float "+o+"() {\n      ivec2 resTexRC = ivec2(resultUV.yx *\n                             vec2("+h[0]+", "+h[1]+"));\n      int index = resTexRC.x * "+h[1]+" + resTexRC.y;\n      "+f+"\n      int texR = index / "+r[1]+";\n      int texC = index - texR * "+r[1]+";\n      vec2 uv = (vec2(texC, texR) + halfCR) /\n                 vec2("+r[1]+".0, "+r[0]+".0);\n\n      return sampleTexture("+i+", uv);\n    }\n  "}function w(e,t){var n=JSON.parse(JSON.stringify(e));return n.shapeInfo.logicalShape=t,n}function x(e,t){return t.map(function(t){return e[t]}).join(", ")}Object.defineProperty(n,"__esModule",{value:!0});var A=e("../../environment"),E=e("../../util"),_=e("../../ops/broadcast_util"),T=e("./tex_util");n.makeShader=function(e,t,n,a){var u=r(),l=i(),c=e.map(function(e){return"uniform sampler2D "+e.name+";"}).join("\n"),p=e.map(function(e){return o(e,t,a)}).join("\n"),h=t.texShape,d=s(t.logicalShape,h);return[R,u,l,c,d,p,n].join("\n")};var S="\n  uniform float NaN;\n\n  const vec4 floatDeltas = vec4(\n      1.0,\n      1.0 / 255.0,\n      1.0 / (255.0 * 255.0),\n      1.0 / (255.0 * 255.0 * 255.0)\n  );\n  const float minValue = "+T.FLOAT_MIN+".0;\n  const float maxValue = "+T.FLOAT_MAX+".0;\n  const float range = (maxValue - minValue) / 255.0;\n  const vec2 dotRange = vec2(1.0, range);\n\n  float sampleTexture(sampler2D textureSampler, vec2 uv) {\n    vec4 sampleValue = texture2D(textureSampler, uv);\n    if (all(equal(sampleValue, vec4("+T.BYTE_NAN_VALUE+")))) {\n      return NaN;\n    }\n\n    vec4 encValue = floor(sampleValue * 255.0 + 0.5);\n    float decodedValue = dot(encValue, floatDeltas);\n    return dot(vec2(minValue, decodedValue), dotRange);\n  }\n",C="\n  const vec4 floatPowers = vec4(\n    1.0,\n    255.0,\n    255.0 * 255.0,\n    255.0 * 255.0 * 255.0\n  );\n  const vec2 recipRange = vec2(1.0/range);\n  const vec2 recipRange255 = vec2(1.0/(maxValue - minValue));\n\n  void setOutput(float decodedValue) {\n    if (isNaN(decodedValue)) {\n      gl_FragColor = vec4("+T.BYTE_NAN_VALUE+");\n      return;\n    }\n\n    float a = dot(vec2(decodedValue, -minValue), recipRange);\n    float b = fract(a) * 255.0;\n    float c = fract(b) * 255.0;\n    float d = fract(c) * 255.0;\n    gl_FragColor = floor(vec4(a, b, c, d)) / 255.0;\n\n    // TODO(dsmilkov): Version above gets better accuracy but probably slower\n    // than the version below. Benchmark to determine if the accuracy is worth\n    // the cost.\n\n    // float normValue = dot(vec2(decodedValue, -minValue), recipRange255);\n    // vec4 f = normValue * floatPowers;\n    // gl_FragColor = floor(fract(f) * 255.0) / 255.0;\n  }\n",O="\n  float sampleTexture(sampler2D textureSampler, vec2 uv) {\n    return texture2D(textureSampler, uv).r;\n  }\n",k="\n  void setOutput(float val) {\n    gl_FragColor = vec4(val, 0, 0, 0);\n  }\n",R="\n  precision highp float;\n  precision highp int;\n  varying vec2 resultUV;\n  const vec2 halfCR = vec2(0.5, 0.5);\n\n  bool isNaN(float val) {\n    float v1 = val * val;\n    float v2 = val * val;\n    return v1 == v2 ? false : true;\n  }\n\n  bool hasNaN(vec4 values) {\n    vec4 v1 = values * values;\n    vec4 v2 = values * values;\n    return any(notEqual(v1, v2));\n  }\n\n  float getNaN(vec4 values) {\n    return dot(vec4(1), values);\n  }\n\n  int round(float value) {\n    return int(floor(value + 0.5));\n  }\n\n  int imod(int x, int y) {\n    return x - y * (x / y);\n  }\n\n  const vec2 randomConst = vec2(\n    23.14069263277926, // e^pi (Gelfond's constant)\n     2.665144142690225 // 2^sqrt(2) (Gelfond–Schneider constant)\n  );\n\n  float random(float seed) {\n      return fract(cos(dot(resultUV * seed, randomConst)) * 12345.6789);\n  }\n\n  \nvec2 UVfrom1D(int texNumR, int texNumC, int index) {\n  int texR = index / texNumC;\n  int texC = index - texR * texNumC;\n  return (vec2(texC, texR) + halfCR) / vec2(texNumC, texNumR);\n}\n\n  \nvec2 UVfrom2D(int texNumR, int texNumC, int numC, int row, int col) {\n  int index = row * numC + col;\n  int texR = index / texNumC;\n  int texC = index - texR * texNumC;\n  return (vec2(texC, texR) + halfCR) / vec2(texNumC, texNumR);\n}\n\n  \nvec2 UVfrom3D(int texNumR, int texNumC, int stride0,\n    int stride1, int row, int col, int depth) {\n  // Explicitly use integer operations as dot() only works on floats.\n  int index = row * stride0 + col * stride1 + depth;\n  int texR = index / texNumC;\n  int texC = index - texR * texNumC;\n  return (vec2(texC, texR) + halfCR) / vec2(texNumC, texNumR);\n}\n\n  \nvec2 UVfrom4D(int texNumR, int texNumC, int stride0,\n    int stride1, int stride2, int row, int col, int depth,\n    int depth2) {\n  // Explicitly use integer operations as dot() only works on floats.\n  int index = row * stride0 + col * stride1 + depth * stride2 + depth2;\n  int texR = index / texNumC;\n  int texC = index - texR * texNumC;\n  return (vec2(texC, texR) + halfCR) / vec2(texNumC, texNumR);\n}\n\n";n.getCoordsDataType=function(e){if(e<=1)return"int";if(2===e)return"ivec2";if(3===e)return"ivec3";if(4===e)return"ivec4";throw Error("GPU for rank "+e+" is not yet supported")}},{"../../environment":5,"../../ops/broadcast_util":50,"../../util":95,"./tex_util":40}],39:[function(e,t,n){"use strict";function r(e){if(1===e)return"sourceLoc";if(2===e)return"sourceLoc.x, sourceLoc.y";if(3===e)return"sourceLoc.x, sourceLoc.y, sourceLoc.z";if(4===e)return"sourceLoc.x, sourceLoc.y, sourceLoc.z, sourceLoc.w";throw Error("Slicing for rank "+e+" is not yet supported")}Object.defineProperty(n,"__esModule",{value:!0});var i=e("./shader_compiler"),a=function(){function e(e){this.variableNames=["source"],this.outputShape=e,this.rank=e.length;var t=i.getCoordsDataType(this.rank),n=r(this.rank);this.userCode="\n      uniform "+t+" start;\n\n      void main() {\n        "+t+" sourceLoc = start + getOutputCoords();\n        setOutput(getSource("+n+"));\n      }\n    "}return e.prototype.getCustomSetupFunc=function(e){var t=this;if(e.length!==this.rank)throw Error("The rank ("+this.rank+") of the program must match the length of start ("+e.length+")");return function(n,r){if(null!=t.startLoc||(t.startLoc=n.getUniformLocationNoThrow(r,"start"),null!=t.startLoc))if(1===t.rank)n.gl.uniform1i(t.startLoc,e[0]);else if(2===t.rank)n.gl.uniform2i(t.startLoc,e[0],e[1]);else if(3===t.rank)n.gl.uniform3i(t.startLoc,e[0],e[1],e[2]);else{if(4!==t.rank)throw Error("Slicing for rank "+t.rank+" is not yet supported");n.gl.uniform4i(t.startLoc,e[0],e[1],e[2],e[3])}}},e}();n.SliceProgram=a},{"./shader_compiler":38}],40:[function(e,t,n){"use strict";function r(e,t){return e*t}function i(e,t){if(e%t!=0)throw new Error("unpackedSize ("+e+") must be a multiple of "+t);return e/t}function a(e,t){return[Math.ceil(t/2),Math.ceil(e/2)]}function o(e,t){var n=a(e,t);return n[0]*n[1]*4}Object.defineProperty(n,"__esModule",{value:!0});!function(e){e[e.FLOAT=0]="FLOAT",e[e.UNSIGNED_BYTE=1]="UNSIGNED_BYTE"}(n.TextureType||(n.TextureType={})),n.getUnpackedMatrixTextureShapeWidthHeight=function(e,t){return[t,e]},n.getUnpackedArraySizeFromMatrixSize=r,n.getColorMatrixTextureShapeWidthHeight=function(e,t){return[4*t,e]},n.getMatrixSizeFromUnpackedArraySize=i,n.encodeMatrixToUnpackedArray=function(e,t,n){var i=r(e.length,n);if(t.length<i)throw new Error("unpackedArray length ("+t.length+") must be >= "+i);for(var a=0,o=0;o<e.length;++o)t[a]=e[o],a+=n},n.FLOAT_MAX=2e4,n.FLOAT_MIN=-n.FLOAT_MAX;var s=(n.FLOAT_MAX-n.FLOAT_MIN)/255,u=[1,1/255,1/65025,1/16581375],l=[1,255,65025];n.BYTE_NAN_VALUE=0,n.encodeFloatArray=function(e){for(var t=new Uint8Array(4*e.length),r=0;r<t.length;r+=4)!function(r){var i=e[r/4];if(isNaN(i))return t[r]=n.BYTE_NAN_VALUE,t[r+1]=n.BYTE_NAN_VALUE,t[r+2]=n.BYTE_NAN_VALUE,t[r+3]=n.BYTE_NAN_VALUE,"continue";var a=(i-n.FLOAT_MIN)/s,o=l.map(function(e){return e*a}).map(function(e){return Math.floor(e%1*255)});t[r]=Math.floor(a),t[r+1]=o[0],t[r+2]=o[1],t[r+3]=o[2]}(r);return t},n.decodeToFloatArray=function(e){for(var t=new Float32Array(e.length/4),r=0;r<e.length;r+=4)!function(r){if(e[r]===n.BYTE_NAN_VALUE&&e[r+1]===n.BYTE_NAN_VALUE&&e[r+2]===n.BYTE_NAN_VALUE&&e[r+3]===n.BYTE_NAN_VALUE)return t[r/4]=NaN,"continue";var i=0;u.forEach(function(t,n){i+=t*e[r+n]});var a=i*s+n.FLOAT_MIN;t[r/4]=a}(r);return t},n.decodeMatrixFromUnpackedArray=function(e,t,n){var r=i(e.length,n);if(t.length<r)throw new Error("matrix length ("+t.length+") must be >= "+r);for(var a=0,o=0;o<e.length;o+=n)t[a++]=e[o]},n.decodeMatrixFromUnpackedColorRGBAArray=function(e,t,n){var r=e.length*n/4;if(t.length<r)throw new Error("matrix length ("+t.length+") must be >= "+r);for(var i=0,a=0;a<e.length;a+=4)for(var o=0;o<n;o++)t[i++]=e[a+o]},n.getPackedMatrixTextureShapeWidthHeight=a,n.getPackedRGBAArraySizeFromMatrixShape=o,n.encodeMatrixToPackedRGBA=function(e,t,n,r){var i=o(t,n);if(r.length<i)throw new Error("packedRGBA length ("+r.length+") must be >= "+i);for(var s=a(t,n),u=s[0],l=s[1],c=n%2==1,p=t%2==1,h=Math.floor(n/2),d=Math.floor(t/2),f=c?4:0,g=n,m=0,v=0;v<d;++v){for(var y=2*v*n,b=0;b<h;++b)w=y+2*b,r[m]=e[w],r[m+1]=e[w+1],r[m+2]=e[w+g],r[m+3]=e[w+g+1],m+=4;m+=f}if(c)for(var w=n-1,m=4*(u-1),x=2*n,f=4*u,v=0;v<d;++v)r[m]=e[w],r[m+2]=e[w+n],w+=x,m+=f;if(p)for(var w=(t-1)*n,m=(l-1)*u*4,b=0;b<h;++b)r[m++]=e[w++],r[m++]=e[w++],m+=2;return c&&p&&(r[r.length-4]=e[e.length-1]),r},n.decodeMatrixFromPackedRGBA=function(e,t,n,r){var i=t*n;if(i<r.length)throw new Error("matrix length ("+r.length+") must be >= "+i);for(var o=n%2==1,s=t%2==1,u=Math.floor(n/2),l=Math.floor(t/2),c=a(t,n),p=c[0],h=c[1],d=o?4:0,f=n+(o?1:0),g=0,m=0,v=n,y=0;y<l;++y){for(w=0;w<u;++w)r[m++]=e[g++],r[m++]=e[g++],r[v++]=e[g++],r[v++]=e[g++];g+=d,m+=f,v+=f}if(o)for(var g=4*(p-1),b=n-1,d=4*p,f=2*n,y=0;y<l;++y)r[b]=e[g],r[b+n]=e[g+2],g+=d,b+=f;if(s)for(var g=(h-1)*p*4,b=(t-1)*n,w=0;w<u;++w)r[b++]=e[g++],r[b++]=e[g++],g+=2;return o&&s&&(r[r.length-1]=e[e.length-4]),r}},{}],41:[function(e,t,n){"use strict";function r(e,t){return e[0]+"_"+e[1]+"_"+t}Object.defineProperty(n,"__esModule",{value:!0});var i=e("./tex_util"),a=function(){function e(e){this.gpgpu=e,this.numUsedTextures=0,this.numFreeTextures=0,this.freeTextures={},this.logEnabled=!1,this.allocatedTextures=[],this.usedTextureCount={}}return e.prototype.acquireTexture=function(e,t){void 0===t&&(t=i.TextureType.FLOAT);var n=r(e,t);if(n in this.freeTextures||(this.freeTextures[n]=[]),n in this.usedTextureCount||(this.usedTextureCount[n]=0),this.usedTextureCount[n]++,this.freeTextures[n].length>0)return this.numFreeTextures--,this.numUsedTextures++,this.log(),this.freeTextures[n].shift();this.numUsedTextures++,this.log();var a=this.gpgpu.createMatrixTexture(e[0],e[1]);return this.allocatedTextures.push(a),a},e.prototype.releaseTexture=function(e,t,n){void 0===n&&(n=i.TextureType.FLOAT);var a=r(t,n);a in this.freeTextures||(this.freeTextures[a]=[]),this.freeTextures[a].push(e),this.numFreeTextures++,this.numUsedTextures--,this.usedTextureCount[a]--,this.log()},e.prototype.log=function(){if(this.logEnabled){var e=this.numFreeTextures+this.numUsedTextures;console.log("Free/Used",this.numFreeTextures+" / "+this.numUsedTextures,"("+e+")")}},e.prototype.getNumUsedTextures=function(){return this.numUsedTextures},e.prototype.getNumFreeTextures=function(){return this.numFreeTextures},e.prototype.dispose=function(){var e=this;null!=this.allocatedTextures&&(this.allocatedTextures.forEach(function(t){e.gpgpu.deleteMatrixTexture(t)}),this.freeTextures=null,this.allocatedTextures=null,this.usedTextureCount=null,this.numUsedTextures=0,this.numFreeTextures=0)},e}();n.TextureManager=a},{"./tex_util":40}],42:[function(e,t,n){"use strict";function r(e){var t=e.length;if(t>4)throw Error("Tile for rank "+t+" is not yet supported");if(1===t)return"imod(resRC, "+e[0]+")";for(var n=["resRC.x","resRC.y","resRC.z","resRC.w"],r=[],i=0;i<e.length;i++)r.push("imod("+n[i]+", "+e[i]+")");return r.join()}Object.defineProperty(n,"__esModule",{value:!0});var i=e("./shader_compiler"),a=function(){return function(e,t){this.variableNames=["A"];for(var n=new Array(e.length),a=0;a<n.length;a++)n[a]=e[a]*t[a];this.outputShape=n,this.rank=n.length;var o=i.getCoordsDataType(this.rank),s=r(e);this.userCode="\n      void main() {\n        "+o+" resRC = getOutputCoords();\n        setOutput(getA("+s+"));\n      }\n    "}}();n.TileProgram=a},{"./shader_compiler":38}],43:[function(e,t,n){"use strict";function r(e){var t=e.length;if(t>4)throw Error("Transpose for rank "+t+" is not yet supported");for(var n=["resRC.x","resRC.y","resRC.z","resRC.w"],r=new Array(t),i=0;i<e.length;i++)r[e[i]]=n[i];return r.join()}Object.defineProperty(n,"__esModule",{value:!0});var i=e("./shader_compiler"),a=function(){return function(e,t){this.variableNames=["A"];for(var n=new Array(e.length),a=0;a<n.length;a++)n[a]=e[t[a]];this.outputShape=n,this.rank=n.length;var o=i.getCoordsDataType(this.rank),s=r(t);this.userCode="\n    void main() {\n      "+o+" resRC = getOutputCoords();\n      setOutput(getA("+s+"));\n    }\n    "}}();n.TransposeProgram=a},{"./shader_compiler":38}],44:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../../ops/selu_util"),i=e("../../ops/erf_util"),a=function(){return function(e,t){this.variableNames=["A"],this.outputShape=e,this.userCode="\n      float unaryOperation(float x) {\n        "+t+"\n      }\n\n      void main() {\n        float x = getAAtOutCoords();\n        float y = unaryOperation(x);\n\n        setOutput(y);\n      }\n    "}}();n.UnaryOpProgram=a;var o="if (isNaN(x)) return x;";n.ABS="return abs(x);",n.RELU="return (x < 0.0) ? 0.0 : x;",n.ELU="return (x >= 0.0) ? x : (exp(x) - 1.0);",n.SELU="\n  // Stable and Attracting Fixed Point (0, 1) for Normalized Weights.\n  // see: https://arxiv.org/abs/1706.02515\n  float scaleAlpha = "+r.SELU_SCALEALPHA+";\n  float scale = "+r.SELU_SCALE+";\n  return (x >= 0.0) ? scale * x : scaleAlpha * (exp(x) - 1.0);\n",n.STEP=function(e){return void 0===e&&(e=0),o+"\n    return x > 0.0 ? 1.0 : float("+e+");\n  "},n.NEG="return -x;",n.CEIL="return ceil(x);",n.FLOOR="return floor(x);",n.SIGN="return sign(x);",n.ROUND="\n  // OpenGL ES does not support round function.\n  // The algorithm is based on banker's rounding.\n  float base = floor(x);\n  if ((x - base) < 0.5) {\n    return floor(x);\n  } else if ((x - base) > 0.5) {\n    return ceil(x);\n  } else {\n    if (mod(base, 2.0) == 0.0) {\n      return base;\n    } else {\n      return base + 1.0;\n    }\n  }\n",n.EXP="return exp(x);",n.EXPM1="return exp(x) - 1.0;",n.LOG="return log(x);",n.LOG1P="return log(1.0 + x);",n.SQRT="return sqrt(x);",n.RSQRT="return inversesqrt(x);",n.SIGMOID="return 1.0 / (1.0 + exp(-1.0 * x));",n.SOFTPLUS="\n  float epsilon = 1.1920928955078125e-7;\n  float threshold = log(epsilon) + 2.0;\n\n  bool too_large = x > -threshold;\n  bool too_small = x < threshold;\n\n  float result;\n  float exp_x = exp(x);\n\n  if (too_large){\n    result = x;\n  }\n  else if (too_small){\n    result = exp_x;\n  }\n  else{\n    result = log(exp_x + 1.0);\n  }\n  return result;\n",n.SIN="return sin(x);",n.COS="return cos(x);",n.TAN="return tan(x);",n.ASIN="return asin(x);",n.ACOS="return acos(x);",n.ATAN=o+"\n  return atan(x);\n",n.SINH="\n  float e2x = exp(x);\n  return (e2x - 1.0 / e2x) / 2.0;\n",n.COSH="\n  float e2x = exp(-x);\n  return (e2x + 1.0 / e2x) / 2.0;\n",n.TANH="\n  float e2x = exp(-2.0 * abs(x));\n  return sign(x) * (1.0 - e2x) / (1.0 + e2x);\n",n.ASINH="return log(x + sqrt(x * x + 1.0));",n.ACOSH="return log(x + sqrt(x * x - 1.0));",n.ATANH="return (log(1.0 + x) - log(1.0 - x)) / 2.0;",n.ERF='\n  // Error function is calculated approximately with elementary function.\n  // See "Handbook of Mathematical Functions with Formulas, \n  // Graphs, and Mathematical Tables", Abramowitz and Stegun.\n  float p = '+i.ERF_P+";\n  float a1 = "+i.ERF_A1+";\n  float a2 = "+i.ERF_A2+";\n  float a3 = "+i.ERF_A3+";\n  float a4 = "+i.ERF_A4+";\n  float a5 = "+i.ERF_A5+";\n  \n  float t = 1.0 / (1.0 + p * x);\n  return 1.0 - (((((a5*t + a4)*t) + a3)*t + a2)*t + a1)*t*exp(-x*x);\n",n.SQUARE="return x * x;",n.RECIPROCAL="return 1.0 / x;",n.LOGICAL_NOT="return float(!(x >= 1.0));",n.TO_INT="return float(int(x));"},{"../../ops/erf_util":56,"../../ops/selu_util":72}],45:[function(e,t,n){"use strict";function r(e,t){var n,r=g.ENV.get("WEBGL_VERSION");if(2===r?n=e.getContext("webgl2",t):1===r&&(n=e.getContext("webgl",t)||e.getContext("experimental-webgl",t)),0===r||null==n)throw new Error("This browser does not support WebGL.");return n}function i(e,t){var n=t();return a(e),n}function a(e){if(m){var t=e.getError();if(t!==e.NO_ERROR)throw new Error("WebGL Error: "+o(e,t))}}function o(e,t){switch(t){case e.NO_ERROR:return"NO_ERROR";case e.INVALID_ENUM:return"INVALID_ENUM";case e.INVALID_VALUE:return"INVALID_VALUE";case e.INVALID_OPERATION:return"INVALID_OPERATION";case e.INVALID_FRAMEBUFFER_OPERATION:return"INVALID_FRAMEBUFFER_OPERATION";case e.OUT_OF_MEMORY:return"OUT_OF_MEMORY";case e.CONTEXT_LOST_WEBGL:return"CONTEXT_LOST_WEBGL";default:return"Unknown error code "+t}}function s(e,t){var n=v.exec(t);if(null==n)return console.log("Couldn't parse line number in error: "+t),void console.log(e);for(var r=+n[1],i=e.split("\n"),a=i.length.toString().length+2,o=i.map(function(e,t){return f.rightPad((t+1).toString(),a)+e}),s=0,u=0;u<o.length;u++)s=Math.max(o[u].length,s);var l=o.slice(0,r-1),c=o.slice(r-1,r),p=o.slice(r);console.log(l.join("\n")),console.log(t.split("\n")[0]),console.log("%c "+f.rightPad(c[0],s),"border:1px solid red; background-color:#e3d2d2; color:#a61717"),console.log(p.join("\n"))}function u(e){return null!=d?d:d=i(e,function(){return e.getParameter(e.MAX_TEXTURE_SIZE)})}function l(e,t,n){h(e,n),i(e,function(){return e.activeTexture(e.TEXTURE0+n)}),i(e,function(){return e.bindTexture(e.TEXTURE_2D,t)})}function c(e,t){switch(t){case e.FRAMEBUFFER_INCOMPLETE_ATTACHMENT:return"FRAMEBUFFER_INCOMPLETE_ATTACHMENT";case e.FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT:return"FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT";case e.FRAMEBUFFER_INCOMPLETE_DIMENSIONS:return"FRAMEBUFFER_INCOMPLETE_DIMENSIONS";case e.FRAMEBUFFER_UNSUPPORTED:return"FRAMEBUFFER_UNSUPPORTED";default:return"unknown error "+t}}function p(e,t,n){var r=i(e,function(){return t()});if(null==r)throw new Error(n);return r}function h(e,t){var n=e.MAX_COMBINED_TEXTURE_IMAGE_UNITS-1,r=t+e.TEXTURE0;if(r<e.TEXTURE0||r>n){var i="[gl.TEXTURE0, gl.TEXTURE"+n+"]";throw new Error("textureUnit must be in "+i+".")}}Object.defineProperty(n,"__esModule",{value:!0});var d=null,f=e("../../util"),g=e("../../environment");n.createWebGLRenderingContext=function(e){var t=document.createElement("canvas");return t.width=1,t.height=1,r(t,e)},n.createWebGLRenderingContextFromCanvas=r,n.callAndCheck=i;var m=!1;n.enableDebugWebGLErrorChecking=function(e){m=e},n.checkWebGLError=a,n.getWebGLErrorMessage=o,n.getExtensionOrThrow=function(e,t){return p(e,function(){return e.getExtension(t)},'Extension "'+t+'" not supported on this browser.')},n.createVertexShader=function(e,t){var n=p(e,function(){return e.createShader(e.VERTEX_SHADER)},"Unable to create vertex WebGLShader.");if(i(e,function(){return e.shaderSource(n,t)}),i(e,function(){return e.compileShader(n)}),!1===e.getShaderParameter(n,e.COMPILE_STATUS))throw console.log(e.getShaderInfoLog(n)),new Error("Failed to compile vertex shader.");return n},n.createFragmentShader=function(e,t){var n=p(e,function(){return e.createShader(e.FRAGMENT_SHADER)},"Unable to create fragment WebGLShader.");if(i(e,function(){return e.shaderSource(n,t)}),i(e,function(){return e.compileShader(n)}),!1===e.getShaderParameter(n,e.COMPILE_STATUS))throw s(t,e.getShaderInfoLog(n)),new Error("Failed to compile fragment shader.");return n};var v=/ERROR: [0-9]+:([0-9]+):/g;n.createProgram=function(e){return p(e,function(){return e.createProgram()},"Unable to create WebGLProgram.")},n.linkProgram=function(e,t){if(i(e,function(){return e.linkProgram(t)}),!1===e.getProgramParameter(t,e.LINK_STATUS))throw console.log(e.getProgramInfoLog(t)),new Error("Failed to link vertex and fragment shaders.")},n.validateProgram=function(e,t){if(i(e,function(){return e.validateProgram(t)}),!1===e.getProgramParameter(t,e.VALIDATE_STATUS))throw console.log(e.getProgramInfoLog(t)),new Error("Shader program validation failed.")},n.createStaticVertexBuffer=function(e,t){var n=p(e,function(){return e.createBuffer()},"Unable to create WebGLBuffer");return i(e,function(){return e.bindBuffer(e.ARRAY_BUFFER,n)}),i(e,function(){return e.bufferData(e.ARRAY_BUFFER,t,e.STATIC_DRAW)}),n},n.createStaticIndexBuffer=function(e,t){var n=p(e,function(){return e.createBuffer()},"Unable to create WebGLBuffer");return i(e,function(){return e.bindBuffer(e.ELEMENT_ARRAY_BUFFER,n)}),i(e,function(){return e.bufferData(e.ELEMENT_ARRAY_BUFFER,t,e.STATIC_DRAW)}),n},n.queryMaxTextureSize=u,n.getChannelsPerTexture=function(){return g.ENV.get("WEBGL_FLOAT_TEXTURE_ENABLED")&&2===g.ENV.get("WEBGL_VERSION")?1:4},n.createTexture=function(e){return p(e,function(){return e.createTexture()},"Unable to create WebGLTexture.")},n.validateTextureSize=function(e,t,n){var r=u(e);if(t<=0||n<=0)throw i="["+t+"x"+n+"]",new Error("Requested texture size "+i+" is invalid.");if(t>r||n>r){var i="["+t+"x"+n+"]",a="["+r+"x"+r+"]";throw new Error("Requested texture size "+i+" greater than WebGL maximum on this browser / GPU "+a+".")}},n.createFramebuffer=function(e){return p(e,function(){return e.createFramebuffer()},"Unable to create WebGLFramebuffer.")},n.bindVertexBufferToProgramAttribute=function(e,t,n,r,a,o,s){var u=e.getAttribLocation(t,n);return-1!==u&&(i(e,function(){return e.bindBuffer(e.ARRAY_BUFFER,r)}),i(e,function(){return e.vertexAttribPointer(u,a,e.FLOAT,!1,o,s)}),i(e,function(){return e.enableVertexAttribArray(u)}),!0)},n.bindTextureUnit=l,n.unbindTextureUnit=function(e,t){h(e,t),i(e,function(){return e.activeTexture(e.TEXTURE0+t)}),i(e,function(){return e.bindTexture(e.TEXTURE_2D,null)})},n.getProgramUniformLocationOrThrow=function(e,t,n){return p(e,function(){return e.getUniformLocation(t,n)},'uniform "'+n+'" not present in program.')},n.getProgramUniformLocation=function(e,t,n){return e.getUniformLocation(t,n)},n.bindTextureToProgramUniformSampler=function(e,t,n,r,a){i(e,function(){return l(e,n,a)}),i(e,function(){return e.uniform1i(r,a)})},n.bindCanvasToFramebuffer=function(e){i(e,function(){return e.bindFramebuffer(e.FRAMEBUFFER,null)}),i(e,function(){return e.viewport(0,0,e.canvas.width,e.canvas.height)}),i(e,function(){return e.scissor(0,0,e.canvas.width,e.canvas.height)})},n.bindColorTextureToFramebuffer=function(e,t,n){i(e,function(){return e.bindFramebuffer(e.FRAMEBUFFER,n)}),i(e,function(){return e.framebufferTexture2D(e.FRAMEBUFFER,e.COLOR_ATTACHMENT0,e.TEXTURE_2D,t,0)})},n.unbindColorTextureFromFramebuffer=function(e,t){i(e,function(){return e.bindFramebuffer(e.FRAMEBUFFER,t)}),i(e,function(){return e.framebufferTexture2D(e.FRAMEBUFFER,e.COLOR_ATTACHMENT0,e.TEXTURE_2D,null,0)})},n.validateFramebuffer=function(e){var t=e.checkFramebufferStatus(e.FRAMEBUFFER);if(t!==e.FRAMEBUFFER_COMPLETE)throw new Error("Error binding framebuffer: "+c(e,t))},n.getFramebufferErrorMessage=c,n.getTextureShapeFromLogicalShape=function(e,t){2!==t.length&&(t=f.squeezeShape(t).newShape);var n=u(e),r=f.sizeFromShape(t);return t.length<=1&&r<=n?[r,1]:2===t.length&&t[0]<=n&&t[1]<=n?t:3===t.length&&t[0]<=n&&t[1]*t[2]<=n?[t[0],t[1]*t[2]]:4===t.length&&t[0]<=n&&t[1]*t[2]*t[3]<=n?[t[0],t[1]*t[2]*t[3]]:f.sizeToSquarishShape(r)}},{"../../environment":5,"../../util":95}],46:[function(e,t,n){"use strict";function r(e,t){if(null==t||"float32"===t)return new Float32Array(e);if("int32"===t)return new Int32Array(e);if("bool"===t)return new Uint8Array(e);throw new Error("Unknown data type $ {dtype}")}function i(e,t){for(var n=r(e,t),i=0;i<n.length;i++)n[i]=1;return n}function a(e,t){return o(e,t)?e:(Array.isArray(e)&&(e=f.flatten(e)),f.copyTypedArray(e,t))}function o(e,t){return e instanceof Float32Array&&"float32"===t||e instanceof Int32Array&&"int32"===t||e instanceof Uint8Array&&"bool"===t}var s=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o},u=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},l=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var c=e("../doc"),p=e("../environment"),h=e("../tensor"),d=e("../tensor_util"),f=e("../util"),g=e("./axis_util"),m=e("./concat"),v=e("./operation"),y=e("./rand"),b=function(){function e(){}return e.tensor=function(e,t,n){void 0===n&&(n="float32");var r=f.inferShape(e);return null!=t&&1!==r.length&&f.assertShapesMatch(t,r,"Error creating a new Tensor. Inferred shape ("+r+") does not match the provided shape ("+t+"). "),f.isTypedArray(e)||Array.isArray(e)||(e=[e]),t=t||r,h.Tensor.make(t,{values:a(e,n)},n)},e.scalar=function(t,n){if(void 0===n&&(n="float32"),f.isTypedArray(t)||Array.isArray(t))throw new Error("Error creating a new Scalar: value must be a primitive (number|boolean)");return e.tensor(t,[],n)},e.tensor1d=function(t,n){void 0===n&&(n="float32");var r=f.inferShape(t);if(1!==r.length)throw new Error("tensor1d() requires values to be a flat/TypedArray");return e.tensor(t,r,n)},e.tensor2d=function(t,n,r){void 0===r&&(r="float32");var i=f.inferShape(t);if(2!==i.length&&1!==i.length)throw new Error("tensor2d() requires values to be number[][] or flat/TypedArray");if(1===i.length&&null==n)throw new Error("tensor2d() requires shape to be provided when `values` are a flat/TypedArray");return n=n||i,e.tensor(t,n,r)},e.tensor3d=function(t,n,r){void 0===r&&(r="float32");var i=f.inferShape(t);if(3!==i.length&&1!==i.length)throw new Error("tensor3d() requires values to be number[][][] or flat/TypedArray");if(1===i.length&&null==n)throw new Error("tensor3d() requires shape to be provided when `values` are a flat array");return n=n||i,e.tensor(t,n,r)},e.tensor4d=function(t,n,r){void 0===r&&(r="float32");var i=f.inferShape(t);if(4!==i.length&&1!==i.length)throw new Error("tensor4d() requires values to be number[][][][] or flat/TypedArray");if(1===i.length&&null==n)throw new Error("tensor4d() requires shape to be provided when `values` are a flat array");return n=n||i,e.tensor(t,n,r)},e.ones=function(e,t){void 0===t&&(t="float32");var n=i(f.sizeFromShape(e),t);return h.Tensor.make(e,{values:n},t)},e.zeros=function(e,t){void 0===t&&(t="float32");var n=r(f.sizeFromShape(e),t);return h.Tensor.make(e,{values:n},t)},e.fill=function(e,t,n){void 0===n&&(n="float32");var r=f.getTypedArrayFromDType(n,f.sizeFromShape(e));return r.fill(t),h.Tensor.make(e,{values:r},n)},e.onesLike=function(t){return f.assertArgumentsAreTensors({x:t},"onesLike"),e.ones(t.shape,t.dtype)},e.zerosLike=function(t){return f.assertArgumentsAreTensors({x:t},"zerosLike"),e.zeros(t.shape,t.dtype)},e.clone=function(e){f.assertArgumentsAreTensors({x:e},"clone");return p.ENV.engine.runKernel(function(t){return h.Tensor.make(e.shape,{dataId:e.dataId},e.dtype)},{x:e},function(e){return{x:function(){return e.toFloat()}}})},e.randomNormal=function(t,n,r,i,a){if(void 0===n&&(n=0),void 0===r&&(r=1),null!=i&&"bool"===i)throw new Error("Unsupported data type "+i);for(var o=new y.MPRandGauss(n,r,i,!1,a),s=e.buffer(t,i),u=0;u<s.values.length;u++)s.values[u]=o.nextValue();return s.toTensor()},e.truncatedNormal=function(t,n,r,i,a){if(void 0===n&&(n=0),void 0===r&&(r=1),null!=i&&"bool"===i)throw new Error("Unsupported data type "+i);for(var o=new y.MPRandGauss(n,r,i,!0,a),s=e.buffer(t,i),u=0;u<s.values.length;u++)s.values[u]=o.nextValue();return s.toTensor()},e.randomUniform=function(t,n,r,i){void 0===n&&(n=0),void 0===r&&(r=1),void 0===i&&(i="float32");for(var a=e.buffer(t,i),o=0;o<a.values.length;o++)a.values[o]=f.randUniform(n,r);return a.toTensor()},e.rand=function(e,t,n){var r=f.sizeFromShape(e),i=null;if(null==n||"float32"===n)i=new Float32Array(r);else if("int32"===n)i=new Int32Array(r);else{if("bool"!==n)throw new Error("Unknown data type "+n);i=new Uint8Array(r)}for(var a=0;a<r;a++)i[a]=t();return h.Tensor.make(e,{values:i},n)},e.multinomial=function(e,t,n,r){void 0===r&&(r=!1),f.assertArgumentsAreTensors({logits:e},"multinomial");var i=e.size,a=e.rank;if(i<2)throw new Error("Error in multinomial: you need at least 2 outcomes, but got "+i+".");if(a>2)throw new Error("Rank of probabilities must be 1 or 2, but is "+a);n=n||Math.random();var o=1===a?e.as2D(1,-1):e,s=p.ENV.engine.runKernel(function(e){return e.multinomial(o,r,t,n)},{logits2D:o});return 1===a?s.as1D():s},e.oneHot=function(e,t,n,r){if(void 0===n&&(n=1),void 0===r&&(r=0),f.assert("int32"===e.dtype,"Indices must be of dtype `int32`"),t<2)throw new Error("Error in oneHot: depth must be >=2, but it is "+t);return p.ENV.engine.runKernel(function(i){return i.oneHot(e,t,n,r)},{indices:e})},e.fromPixels=function(e,t){if(void 0===t&&(t=3),t>4)throw new Error("Cannot construct Tensor with more than 4 channels from pixels.");return p.ENV.engine.fromPixels(e,t)},e.toPixels=function(e,t){return u(this,void 0,void 0,function(){var n,r,i,a,o,s,u,c,p,h,d,g,m,v,y,b,w;return l(this,function(l){switch(l.label){case 0:if(f.assertArgumentsAreTensors({img:e},"toPixels"),2!==e.rank&&3!==e.rank)throw new Error("toPixels only supports rank 2 or 3 tensors, got rank "+e.rank+".");if(n=e.shape.slice(0,2),r=n[0],i=n[1],(a=2===e.rank?1:e.shape[2])>4||2===a)throw new Error("toPixels only supports depth of size 1, 3 or 4 but got "+a);return[4,e.min().data()];case 1:return o=l.sent()[0],[4,e.max().data()];case 2:if(s=l.sent()[0],"float32"===e.dtype){if(o<0||s>1)throw new Error("Tensor values for a float32 Tensor must be in the range [0 - 1] but got range ["+o+" - "+s+"].")}else{if("int32"!==e.dtype)throw new Error("Unsupported type for toPixels: "+e.dtype+". Please use float32 or int32 tensors.");if(o<0||s>255)throw new Error("Tensor values for a int32 Tensor must be in the range [0 - 255] but got range ["+o+" - "+s+"].")}return[4,e.data()];case 3:for(u=l.sent(),c="float32"===e.dtype?255:1,p=new Uint8ClampedArray(i*r*4),h=0;h<r*i;++h)d=void 0,g=void 0,m=void 0,v=void 0,1===a?(d=u[h]*c,g=u[h]*c,m=u[h]*c,v=255):3===a?(d=u[3*h]*c,g=u[3*h+1]*c,m=u[3*h+2]*c,v=255):4===a&&(d=u[4*h]*c,g=u[4*h+1]*c,m=u[4*h+2]*c,v=u[4*h+3]*c),p[(y=4*h)+0]=Math.round(d),p[y+1]=Math.round(g),p[y+2]=Math.round(m),p[y+3]=Math.round(v);return null!=t&&(t.width=i,t.height=r,b=t.getContext("2d"),w=new ImageData(p,i,r),b.putImageData(w,0,0)),[2,p]}})})},e.reshape=function(e,t){f.assertArgumentsAreTensors({x:e},"reshape"),t=f.inferFromImplicitShape(t,e.size),f.assert(e.size===f.sizeFromShape(t),"new shape and old shape must have the same number of elements.");return p.ENV.engine.runKernel(function(n){return n.reshape(e,t)},{x:e},function(t){return{x:function(){return t.reshape(e.shape)}}})},e.squeeze=function(t,n){return f.assertArgumentsAreTensors({x:t},"squeeze"),e.reshape(t,f.squeezeShape(t.shape,n).newShape)},e.cast=function(e,t){f.assertArgumentsAreTensors({x:e},"cast");return p.ENV.engine.runKernel(function(n){return n.cast(e,t)},{x:e},function(e){return{x:function(){return e.clone()}}})},e.tile=function(t,n){f.assertArgumentsAreTensors({x:t},"tile"),f.assert(t.rank===n.length,"Error in transpose: rank of input "+t.rank+" must match length of reps "+n+".");return p.ENV.engine.runKernel(function(e){return e.tile(t,n)},{x:t},function(r){return{x:function(){var i=e.zerosLike(t);if(1===t.rank)for(a=0;a<n[0];++a)i=i.add(r.slice([a*t.shape[0]],[t.shape[0]]));else if(2===t.rank)for(a=0;a<n[0];++a)for(o=0;o<n[1];++o)i=i.add(r.slice([a*t.shape[0],o*t.shape[1]],[t.shape[0],t.shape[1]]));else if(3===t.rank)for(a=0;a<n[0];++a)for(o=0;o<n[1];++o)for(s=0;s<n[2];++s)i=i.add(r.slice([a*t.shape[0],o*t.shape[1],s*t.shape[2]],[t.shape[0],t.shape[1],t.shape[2]]));else{if(4!==t.rank)throw new Error("Gradient for tile operation is not implemented for rank-"+t.rank+" tensors yet.");for(var a=0;a<n[0];++a)for(var o=0;o<n[1];++o)for(var s=0;s<n[2];++s)for(var u=0;u<n[3];++u)i=i.add(r.slice([a*t.shape[0],o*t.shape[1],s*t.shape[2],u*t.shape[3]],[t.shape[0],t.shape[1],t.shape[2],t.shape[3]]))}return i}}})},e.gather=function(e,t,n){void 0===n&&(n=0),f.assertArgumentsAreTensors({x:e,indices:t},"gather"),f.assert("int32"===t.dtype,"Indices must be of dtype `int32`");var r=g.parseAxisParam(n,e.shape);return p.ENV.engine.runKernel(function(n){return n.gather(e,t,r[0])},{x:e,indices:t})},e.pad1d=function(t,n,r){return void 0===r&&(r=0),f.assert(2===n.length,"Invalid number of paddings. Must be length of 2."),e.pad(t,[n],r)},e.pad2d=function(t,n,r){return void 0===r&&(r=0),f.assert(2===n.length&&2===n[0].length&&2===n[1].length,"Invalid number of paddings. Must be length of 2 each."),e.pad(t,n,r)},e.pad3d=function(t,n,r){return void 0===r&&(r=0),f.assert(3===n.length&&2===n[0].length&&2===n[1].length&&2===n[2].length,"Invalid number of paddings. Must be length of 2 each."),e.pad(t,n,r)},e.pad4d=function(t,n,r){return void 0===r&&(r=0),f.assert(4===n.length&&2===n[0].length&&2===n[1].length&&2===n[2].length&&2===n[3].length,"Invalid number of paddings. Must be length of 2 each."),e.pad(t,n,r)},e.pad=function(e,t,n){if(void 0===n&&(n=0),f.assertArgumentsAreTensors({x:e},"pad"),0===e.rank)throw new Error("pad(scalar) is not defined. Pass non-scalar to pad");var r=t.map(function(e){return e[0]});return p.ENV.engine.runKernel(function(r){return r.pad(e,t,n)},{x:e},function(t){return{x:function(){return t.slice(r,e.shape)}}})},e.stack=function(e,t){if(void 0===t&&(t=0),f.assertArgumentsAreTensors({tensors:e},"stack"),f.assert(e.length>=1,"Pass at least one tensor to tf.stack"),1===e.length)return e[0].expandDims(t);var n=e[0].rank,r=e[0].shape,i=e[0].dtype;f.assert(t<=n,"Axis must be <= rank of the tensor"),e.forEach(function(e){f.assertShapesMatch(r,e.shape,"All tensors passed to stack must have matching shapes")}),e.forEach(function(e){f.assert(i===e.dtype,"All tensors passed to stack must have matching dtypes")});var a=e.map(function(e){return e.expandDims(t)});return m.ConcatOps.concat(a,t)},e.split=function(e,t,n){void 0===n&&(n=0),f.assertArgumentsAreTensors({x:e},"split"),n=g.parseAxisParam(n,e.shape)[0];var r;"number"==typeof t?(f.assert(e.shape[n]%t==0,"Number of splits must evenly divide the axis."),r=Array(t).fill(e.shape[n]/t)):(f.assert(e.shape[n]===t.reduce(function(e,t){return e+t}),"The sum of sizes must match the size of the axis dimension."),r=t);var i=Array(e.rank).fill(0),a=e.shape.slice();return r.map(function(t){a[n]=t;var r=e.slice(i,a);return i[n]+=t,r})},e.expandDims=function(t,n){void 0===n&&(n=0),f.assertArgumentsAreTensors({x:t},"expandDims"),f.assert(n<=t.rank,"Axis must be <= rank of the tensor");var r=t.shape.slice();return r.splice(n,0,1),e.reshape(t,r)},e.linspace=function(t,n,i){if(0===i)throw new Error("Cannot request zero samples");var a=(n-t)/(i-1),o=r(i,"float32");o[0]=t;for(var s=1;s<o.length;s++)o[s]=o[s-1]+a;return e.tensor1d(o,"float32")},e.range=function(t,n,i,a){if(void 0===i&&(i=1),void 0===a&&(a="float32"),0===i)throw new Error("Cannot have a step of zero");var o=t===n,s=t<n&&i<0,u=n<t&&i>1;if(o||s||u)return e.zeros([0],a);var l=r(Math.abs(Math.ceil((n-t)/i)),a);n<t&&1===i&&(i=-1),l[0]=t;for(var c=1;c<l.length;c++)l[c]=l[c-1]+i;return e.tensor1d(l,a)},e.buffer=function(e,t,n){return void 0===t&&(t="float32"),new h.TensorBuffer(e,t,n)},e.print=function(e,t){void 0===t&&(t=!1),console.log(d.tensorToString(e,t))},s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"tensor",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"scalar",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"tensor1d",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"tensor2d",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"tensor3d",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"tensor4d",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"ones",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"zeros",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"fill",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"onesLike",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"zerosLike",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"clone",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"randomNormal",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"truncatedNormal",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"randomUniform",null),s([v.operation],e,"rand",null),s([v.operation],e,"multinomial",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"oneHot",null),s([c.doc({heading:"Tensors",subheading:"Creation"}),v.operation],e,"fromPixels",null),s([c.doc({heading:"Visualization"})],e,"toPixels",null),s([c.doc({heading:"Tensors",subheading:"Transformations"}),v.operation],e,"reshape",null),s([c.doc({heading:"Tensors",subheading:"Transformations"})],e,"squeeze",null),s([c.doc({heading:"Tensors",subheading:"Transformations"}),v.operation],e,"cast",null),s([c.doc({heading:"Tensors",subheading:"Slicing and Joining"}),v.operation],e,"tile",null),s([c.doc({heading:"Tensors",subheading:"Slicing and Joining"}),v.operation],e,"gather",null),s([c.doc({heading:"Tensors",subheading:"Transformations"}),v.operation],e,"pad",null),s([c.doc({heading:"Tensors",subheading:"Slicing and Joining"}),v.operation],e,"stack",null),s([c.doc({heading:"Tensors",subheading:"Slicing and Joining"}),v.operation],e,"split",null),s([c.doc({heading:"Tensors",subheading:"Transformations"}),v.operation],e,"expandDims",null),s([v.operation,c.doc({heading:"Tensors",subheading:"Creation"})],e,"linspace",null),s([v.operation,c.doc({heading:"Tensors",subheading:"Creation"})],e,"range",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"buffer",null),s([c.doc({heading:"Tensors",subheading:"Creation"})],e,"print",null),e}();n.ArrayOps=b},{"../doc":3,"../environment":5,"../tensor":89,"../tensor_util":90,"../util":95,"./axis_util":47,"./concat":52,"./operation":65,"./rand":68}],47:[function(e,t,n){"use strict";function r(e,t){for(var n=0;n<e.length;++n)if(e[e.length-n-1]!==t-1-n)return!1;return!0}function i(e,t,n){for(var r=e.length+t.length,i=[],a=0,o=0,s=0;s<r;s++)-1===n.indexOf(s)?i.push(e[a++]):i.push(t[o++]);return i}Object.defineProperty(n,"__esModule",{value:!0});var a=e("../util");n.axesAreInnerMostDims=r,n.combineLocations=i,n.computeOutAndReduceShapes=function(e,t){for(var n=[],r=e.length,i=0;i<r;i++)-1===t.indexOf(i)&&n.push(e[i]);return[n,t.map(function(t){return e[t]})]},n.expandShapeToKeepDim=function(e,t){return i(e,t.map(function(e){return 1}),t)},n.parseAxisParam=function(e,t){var n=t.length;return e=null==e?t.map(function(e,t){return t}):[].concat(e),a.assert(e.every(function(e){return e>=-n&&e<n}),"All values in axis param must be in range [-"+n+", "+n+") but got axis "+e),a.assert(e.every(function(e){return a.isInt(e)}),"All values in axis param must be integers but got axis "+e),e.map(function(e){return e<0?n+e:e})},n.assertAxesAreInnerMostDims=function(e,t,n){a.assert(r(t,n),e+" supports only inner-most axes for now. Got axes "+t+" and rank-"+n+" input.")},n.getAxesPermutation=function(e,t){if(r(e,t))return null;for(var n=[],i=0;i<t;++i)-1===e.indexOf(i)&&n.push(i);return e.forEach(function(e){return n.push(e)}),n},n.getUndoAxesPermutation=function(e){return e.map(function(e,t){return[t,e]}).sort(function(e,t){return e[1]-t[1]}).map(function(e){return e[0]})},n.getInnerMostAxes=function(e,t){for(var n=[],r=t-e;r<t;++r)n.push(r);return n}},{"../util":95}],48:[function(e,t,n){"use strict";function r(e){return null==e?null:0===e.rank?e.as1D():1===e.rank?e:2===e.rank?e.as4D(1,1,e.shape[0],e.shape[1]):3===e.rank?e.as4D(1,e.shape[0],e.shape[1],e.shape[2]):e}var i=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var a=e("../doc"),o=e("../environment"),s=e("../util"),u=e("./array_ops"),l=e("./broadcast_util"),c=e("./operation"),p=e("./ops"),h=function(){function e(){}return e.batchNormalization2d=function(t,n,r,i,a,o){return void 0===i&&(i=.001),s.assert(2===t.rank,"Error in batchNormalization3D: x must be rank 3 but got rank "+t.rank+"."),s.assert(2===n.rank||1===n.rank,"Error in batchNormalization2D: mean must be rank 2 or rank 1 but got rank "+n.rank+"."),s.assert(2===r.rank||1===r.rank,"Error in batchNormalization2D: variance must be rank 2 or rank 1 but got rank "+r.rank+"."),null!=a&&s.assert(2===a.rank||1===a.rank,"Error in batchNormalization2D: scale must be rank 2 or rank 1 but got rank "+a.rank+"."),null!=o&&s.assert(2===o.rank||1===o.rank,"Error in batchNormalization2D: offset must be rank 2 or rank 1 but got rank "+o.rank+"."),e.batchNormalization(t,n,r,i,a,o)},e.batchNormalization3d=function(t,n,r,i,a,o){return void 0===i&&(i=.001),s.assert(3===t.rank,"Error in batchNormalization3D: x must be rank 3 but got rank "+t.rank+"."),s.assert(3===n.rank||1===n.rank,"Error in batchNormalization3D: mean must be rank 3 or rank 1 but got rank "+n.rank+"."),s.assert(3===r.rank||1===r.rank,"Error in batchNormalization3D: variance must be rank 3 or rank 1 but got rank "+r.rank+"."),null!=a&&s.assert(3===a.rank||1===a.rank,"Error in batchNormalization3D: scale must be rank 3 or rank 1 but got rank "+a.rank+"."),null!=o&&s.assert(3===o.rank||1===o.rank,"Error in batchNormalization3D: offset must be rank 3 or rank 1 but got rank "+o.rank+"."),e.batchNormalization(t,n,r,i,a,o)},e.batchNormalization4d=function(t,n,r,i,a,o){return void 0===i&&(i=.001),s.assert(4===t.rank,"Error in batchNormalization4D: x must be rank 4 but got rank "+t.rank+"."),s.assert(4===n.rank||1===n.rank,"Error in batchNormalization4D: mean must be rank 4 or rank 1 but got rank "+n.rank+"."),s.assert(4===r.rank||1===r.rank,"Error in batchNormalization4D: variance must be rank 4 or rank 1 but got rank "+r.rank+"."),null!=a&&s.assert(4===a.rank||1===a.rank,"Error in batchNormalization4D: scale must be rank 4 or rank 1 but got rank "+a.rank+"."),null!=o&&s.assert(4===o.rank||1===o.rank,"Error in batchNormalization4D: offset must be rank 4 or rank 1 but got rank "+o.rank+"."),e.batchNormalization(t,n,r,i,a,o)},e.batchNormalization=function(e,t,n,i,a,c){void 0===i&&(i=.001),s.assertArgumentsAreTensors({x:e,mean:t,variance:n},"batchNormalization"),null!=a&&s.assertArgumentsAreTensors({scale:a},"batchNormalization"),null!=c&&s.assertArgumentsAreTensors({offset:c},"batchNormalization"),s.assert(t.rank===n.rank,"Batch normalization gradient requires mean and variance to have equal ranks."),s.assert(null==c||t.rank===c.rank,"Batch normalization gradient requires mean and offset to have equal ranks."),s.assert(null==a||t.rank===a.rank,"Batch normalization gradient requires mean and scale to have equal ranks.");var h;h=0===e.rank||1===e.rank?e.as4D(1,1,1,e.size):2===e.rank?e.as4D(1,1,e.shape[0],e.shape[1]):3===e.rank?e.as4D(1,e.shape[0],e.shape[1],e.shape[2]):e;return o.ENV.engine.runKernel(function(e){return e.batchNormalization(h,r(t),r(n),i,r(a),r(c))},{x:e,mean:t,variance:n,scale:a,offset:c},function(r){var o=null==a?u.ArrayOps.scalar(1):a,s=l.getReductionAxes(t.shape,h.shape),c=[];if(1===t.rank){for(var d=0;d<h.shape.length-1;++d)c.push(h.shape[d]);c.push(1)}var f=e.sub(t),g=r.mul(o),m=p.rsqrt(n.add(u.ArrayOps.scalar(i))),v=m.mul(m).mul(m).mul(u.ArrayOps.scalar(-.5));return{x:function(){return 1===t.rank?r.mul(u.ArrayOps.tile(m.as4D(1,1,1,t.shape[0]),c)).mul(o).reshape(e.shape):r.mul(m).mul(o).reshape(e.shape)},mean:function(){var e=m.mul(u.ArrayOps.scalar(-1)).mul(g);return 1===t.rank&&(e=e.sum(s)),e.reshape(t.shape)},variance:function(){var e=v.mul(f).mul(g);return 1===t.rank&&(e=e.sum(s)),e.reshape(t.shape)},scale:function(){var e=f.mul(m),n=r.mul(e);return 1===t.rank&&(n=n.sum(s)),n.reshape(t.shape)},offset:function(){var e=r;return 1===t.rank&&(e=e.sum(s)),e.reshape(t.shape)}}}).reshape(e.shape)},i([c.operation],e,"batchNormalization2d",null),i([c.operation],e,"batchNormalization3d",null),i([c.operation],e,"batchNormalization4d",null),i([a.doc({heading:"Operations",subheading:"Normalization"})],e,"batchNormalization",null),e}();n.BatchNormOps=h},{"../doc":3,"../environment":5,"../util":95,"./array_ops":46,"./broadcast_util":50,"./operation":65,"./ops":66}],49:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../types"),s=e("../util"),u=e("./broadcast_util"),l=e("./operation"),c=e("./ops"),p=function(){function e(){}return e.add=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"add"),s.assertTypesMatch(e,t);var n=u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.add(e,t)},{a:e,b:t},function(r){return{a:function(){var t=r,i=u.getReductionAxes(e.shape,n);return i.length>0&&(t=t.sum(i)),t.reshape(e.shape)},b:function(){var e=r,i=u.getReductionAxes(t.shape,n);return i.length>0&&(e=e.sum(i)),e.reshape(t.shape)}}})},e.addStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in addStrict: "),e.add(t)},e.sub=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"sub"),s.assertTypesMatch(e,t);var n=u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.subtract(e,t)},{a:e,b:t},function(r){return{a:function(){var t=r,i=u.getReductionAxes(e.shape,n);return i.length>0&&(t=t.sum(i)),t.reshape(e.shape)},b:function(){var e=r,i=u.getReductionAxes(t.shape,n);return i.length>0&&(e=e.sum(i)),e.neg().reshape(t.shape)}}})},e.subStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in subStrict: "),e.sub(t)},e.pow=function(e,t){s.assertArgumentsAreTensors({base:e,exp:t},"pow");var n=u.assertAndGetBroadcastShape(e.shape,t.shape);e=e.cast(o.upcastType(e.dtype,t.dtype)),t=t.cast(o.upcastType(e.dtype,t.dtype));return a.ENV.engine.runKernel(function(n,r){return r(n.pow(e,t))},{base:e,exp:t},function(r,i){var a=i[0];return{base:function(){var i=r.mul(t.toFloat().mul(a.div(e))),o=u.getReductionAxes(e.shape,n);return o.length>0&&(i=i.sum(o)),i.reshape(e.shape)},exp:function(){var i=r.mul(a.mul(e.log()).toFloat()),o=u.getReductionAxes(t.shape,n);return o.length>0&&(i=i.sum(o)),i.reshape(t.shape)}}})},e.powStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in powStrict: "),e.pow(t)},e.mul=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"mul"),s.assertTypesMatch(e,t);var n=u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.multiply(e,t)},{a:e,b:t},function(r){return{a:function(){var i=r.mul(t.toFloat()),a=u.getReductionAxes(e.shape,n);return a.length>0?i.sum(a).reshape(e.shape):i},b:function(){var i=r.mul(e.toFloat()),a=u.getReductionAxes(t.shape,n);return a.length>0?i.sum(a).reshape(t.shape):i}}})},e.mulStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in multiplyStrict: "),e.mul(t)},e.div=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"div"),s.assertTypesMatch(e,t);var n=u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.divide(e,t)},{a:e,b:t},function(r){return{a:function(){var i=r.div(t.toFloat()),a=u.getReductionAxes(e.shape,n);return a.length>0?i.sum(a).reshape(e.shape):i},b:function(){var i=r.mul(e.toFloat()),a=u.getReductionAxes(t.shape,n);a.length>0&&(i=i.sum(a).reshape(t.shape));var o=t.square();return i.div(o.toFloat()).neg()}}})},e.divStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in divideStrict: "),e.div(t)},e.mod=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"mod"),s.assertTypesMatch(e,t);var n=u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.mod(e,t)},{a:e,b:t},function(r){return{a:function(){var t=u.getReductionAxes(e.shape,n);return t.length>0?r.sum(t).reshape(e.shape):r},b:function(){var i=r.mul(e.div(t).floor().neg()),a=u.getReductionAxes(t.shape,n);return a.length>0?i.sum(a).reshape(t.shape):i}}})},e.modStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in modStrict: "),e.mod(t)},e.minimum=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"minimum"),s.assertTypesMatch(e,t),"bool"===e.dtype&&(e=e.toInt()),"bool"===t.dtype&&(t=t.toInt()),u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.minimum(e,t)},{a:e,b:t},function(n){return{a:function(){return n.mul(e.lessEqual(t).toFloat())},b:function(){return n.mul(e.greater(t).toFloat())}}})},e.minimumStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in minimumStrict: "),e.minimum(t)},e.maximum=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"maximum"),s.assertTypesMatch(e,t),"bool"===e.dtype&&(e=e.toInt()),"bool"===t.dtype&&(t=t.toInt()),u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.maximum(e,t)},{a:e,b:t},function(n){return{a:function(){return n.mul(e.greaterEqual(t).toFloat())},b:function(){return n.mul(e.less(t).toFloat())}}})},e.maximumStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in minimumStrict: "),e.maximum(t)},e.squaredDifference=function(e,t){s.assertArgumentsAreTensors({a:e,b:t},"squaredDifference"),s.assertTypesMatch(e,t),u.assertAndGetBroadcastShape(e.shape,t.shape);return a.ENV.engine.runKernel(function(n){return n.squaredDifference(e,t)},{a:e,b:t},function(n){var r=c.scalar(2);return{a:function(){return n.mul(e.sub(t).mul(r))},b:function(){return n.mul(t.sub(e).mul(r))}}})},e.squaredDifferenceStrict=function(e,t){return s.assertShapesMatch(e.shape,t.shape,"Error in squaredDifferenceStrict: "),e.squaredDifference(t)},e.atan2=function(t,n){s.assertArgumentsAreTensors({a:t,b:n},"atan2"),s.assertTypesMatch(t,n);var r=u.assertAndGetBroadcastShape(t.shape,n.shape);return a.ENV.engine.runKernel(function(e){return e.atan2(t,n)},{a:t,b:n},function(i){return{a:function(){var a=e.add(c.square(t),c.square(n)),o=i.mul(n.div(a)),s=u.getReductionAxes(t.shape,r);return s.length>0&&(o=o.sum(s)),o.reshape(t.shape)},b:function(){var a=e.add(c.square(t),c.square(n)),o=c.neg(i.mul(t.div(a))),s=u.getReductionAxes(n.shape,r);return s.length>0&&(o=o.sum(s)),o.reshape(n.shape)}}})},r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"add",null),r([l.operation],e,"addStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"sub",null),r([l.operation],e,"subStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"pow",null),r([l.operation],e,"powStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"mul",null),r([l.operation],e,"mulStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"div",null),r([l.operation],e,"divStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"mod",null),r([l.operation],e,"modStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"minimum",null),r([l.operation],e,"minimumStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"maximum",null),r([l.operation],e,"maximumStrict",null),r([i.doc({heading:"Operations",subheading:"Arithmetic"}),l.operation],e,"squaredDifference",null),r([l.operation],e,"squaredDifferenceStrict",null),r([l.operation],e,"atan2",null),e}();n.BinaryOps=p},{"../doc":3,"../environment":5,"../types":94,"../util":95,"./broadcast_util":50,"./operation":65,"./ops":66}],50:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0}),n.getBroadcastDims=function(e,t){for(var n=e.length,r=[],i=0;i<n;i++){var a=n-1-i,o=e[a]||1;(t[t.length-1-i]||1)>1&&1===o&&r.unshift(a)}return r},n.getReductionAxes=function(e,t){for(var n=[],r=0;r<t.length;r++){var i=e[e.length-r-1],a=t.length-r-1,o=t[a];(null==i||1===i&&o>1)&&n.unshift(a)}return n},n.broadcastDimsAreOuter=function(e){for(var t=0;t<e.length;t++)if(e[t]!==t)return!1;return!0},n.assertAndGetBroadcastShape=function(e,t){for(var n=[],r="Operands could not be broadcast together with shapes "+e+" and "+t+".",i=Math.max(e.length,t.length),a=0;a<i;a++){var o=e[e.length-a-1]||1,s=t[t.length-a-1]||1;if(o>1&&s>1&&o!==s)throw Error(r);n.unshift(Math.max(o,s))}return n}},{}],51:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./broadcast_util"),u=e("./operation"),l=function(){function e(){}return e.notEqual=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"notEqual"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.notEqual(e,t)},{a:e,b:t})},e.notEqualStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in notEqualStrict: "),e.notEqual(t)},e.less=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"less"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.less(e,t)},{a:e,b:t})},e.lessStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in lessStrict: "),e.less(t)},e.equal=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"equal"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.equal(e,t)},{a:e,b:t})},e.equalStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in equalStrict: "),e.equal(t)},e.lessEqual=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"lessEqual"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.lessEqual(e,t)},{a:e,b:t})},e.lessEqualStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in lessEqualStrict: "),e.lessEqual(t)},e.greater=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"greater"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.greater(e,t)},{a:e,b:t})},e.greaterStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in greaterStrict: "),e.greater(t)},e.greaterEqual=function(e,t){return o.assertArgumentsAreTensors({a:e,b:t},"greaterEqual"),o.assertTypesMatch(e,t),s.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.greaterEqual(e,t)},{a:e,b:t})},e.greaterEqualStrict=function(e,t){return o.assertShapesMatch(e.shape,t.shape,"Error in greaterEqualStrict: "),e.greaterEqual(t)},r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"notEqual",null),r([u.operation],e,"notEqualStrict",null),r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"less",null),r([u.operation],e,"lessStrict",null),r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"equal",null),r([u.operation],e,"equalStrict",null),r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"lessEqual",null),r([u.operation],e,"lessEqualStrict",null),r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"greater",null),r([u.operation],e,"greaterStrict",null),r([i.doc({heading:"Operations",subheading:"Logical"}),u.operation],e,"greaterEqual",null),r([u.operation],e,"greaterEqualStrict",null),e}();n.CompareOps=l},{"../doc":3,"../environment":5,"../util":95,"./broadcast_util":50,"./operation":65}],52:[function(e,t,n){"use strict";function r(e,t,n){l.assertParams(e.shape,t.shape,n);var r=l.computeOutShape(e.shape,t.shape,n),i=e.as2D(-1,s.sizeFromShape(e.shape.slice(n))),a=t.as2D(-1,s.sizeFromShape(t.shape.slice(n))),u=l.computeGradientSliceShapes(i.shape,a.shape),c=u.aBegin,p=u.aSize,h=u.bBegin,d=u.bSize;return o.ENV.engine.runKernel(function(e){return e.concat(i,a)},{a:i,b:a},function(e){return{a:function(){return e.slice(c,p)},b:function(){return e.slice(h,d)}}}).reshape(r)}var i=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var a=e("../doc"),o=e("../environment"),s=e("../util"),u=e("./axis_util"),l=e("./concat_util"),c=e("./operation"),p=function(){function e(){}return e.concat1d=function(t){return e.concat(t,0)},e.concat2d=function(t,n){return e.concat(t,n)},e.concat3d=function(t,n){return e.concat(t,n)},e.concat4d=function(t,n){return e.concat(t,n)},e.concat=function(e,t){void 0===t&&(t=0),s.assert(e.length>=1,"Pass at least one tensor to concat"),s.assertArgumentsAreTensors({tensors:e},"concat");var n=e[0];if(1===e.length)return n;for(var i=u.parseAxisParam(t,n.shape),a=1;a<e.length;++a)n=r(n,e[a],i[0]);return n},i([a.doc({heading:"Tensors",subheading:"Slicing and Joining"}),c.operation],e,"concat",null),e}();n.ConcatOps=p},{"../doc":3,"../environment":5,"../util":95,"./axis_util":47,"./concat_util":53,"./operation":65}],53:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../util");n.assertParams=function(e,t,n){var i=e.length,a=t.length;r.assert(e.length===t.length,"Error in concat"+i+"D: rank of x1 ("+i+") and x2 ("+a+") must be the same."),r.assert(n>=0&&n<i,"Error in concat"+i+"D: axis must be between 0 and "+(i-1)+".");for(var o=0;o<i;o++)r.assert(o===n||e[o]===t[o],"Error in concat"+i+"D: Shape ("+e+") does not match ("+t+") along the non-concatenated axis "+o+".")},n.computeOutShape1D=function(e,t){r.assert(1===e.length&&1===t.length,"x1 and x2 should be 1d array.");var n=e.slice();return n[0]+=t[0],n},n.computeOutShape=function(e,t,n){r.assert(e.length===t.length,"x1 and x2 should have the same rank.");var i=e.slice();return i[n]+=t[n],i},n.computeGradientSliceShapes=function(e,t){return{aBegin:[0,0],aSize:e,bBegin:[0,e[1]],bSize:t}}},{"../util":95}],54:[function(e,t,n){"use strict";function r(e){return"number"==typeof e?[e,e]:e}function i(e){var t=r(e),n=t[0],i=t[1];return 1===n&&1===i}function a(e,t){return i(e)||i(t)}var o=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var s=e("../doc"),u=e("../environment"),l=e("../util"),c=e("./conv_util"),p=e("./operation"),h=function(){function e(){}return e.conv1d=function(t,n,r,i,o,s,u){void 0===o&&(o="NWC"),void 0===s&&(s=1),l.assertArgumentsAreTensors({x:t,filter:n},"conv1d");var c=t,p=!1;2===t.rank&&(p=!0,c=t.as3D(1,t.shape[0],t.shape[1])),l.assert(3===c.rank,"Error in conv1d: input must be rank 3, but got rank "+c.rank+"."),l.assert(3===n.rank,"Error in conv1d: filter must be rank 3, but got rank "+n.rank+"."),null!=u&&l.assert(l.isInt(i),"Error in conv1d: pad must be an integer when using, dimRoundingMode "+u+" but got pad "+i+"."),l.assert(c.shape[2]===n.shape[1],"Error in conv1d: depth of input ("+c.shape[2]+") must match  input depth for filter "+n.shape[1]+"."),l.assert(a(r,s),"Error in conv1D: Either stride or dilation must be 1.Got stride "+r+" and dilation '"+s+"'"),l.assert("NWC"===o,"Error in conv1d: got dataFormat of "+o+" but only NWC is currently supported.");var h=n.as4D(1,n.shape[0],n.shape[1],n.shape[2]),d=c.as4D(c.shape[0],1,c.shape[1],c.shape[2]),f=[1,r],g=[1,s],m=e.conv2d(d,h,f,i,"NHWC",g,u);return p?m.as2D(m.shape[2],m.shape[3]):m.as3D(m.shape[0],m.shape[2],m.shape[3])},e.conv2d=function(t,n,r,o,s,p,h){void 0===s&&(s="NHWC"),void 0===p&&(p=[1,1]),l.assertArgumentsAreTensors({x:t,filter:n},"conv2d");var d=t,f=!1;3===t.rank&&(f=!0,d=t.as4D(1,t.shape[0],t.shape[1],t.shape[2])),l.assert(4===d.rank,"Error in conv2d: input must be rank 4, but got rank "+d.rank+"."),l.assert(4===n.rank,"Error in conv2d: filter must be rank 4, but got rank "+n.rank+"."),null!=h&&l.assert(l.isInt(o),"Error in conv2d: pad must be an integer when using, dimRoundingMode "+h+" but got pad "+o+"."),l.assert(d.shape[3]===n.shape[2],"Error in conv2d: depth of input ("+d.shape[3]+") must match  input depth for filter "+n.shape[2]+"."),l.assert(a(r,p),"Error in conv2D: Either strides or dilations must be 1.Got strides "+r+" and dilations '"+p+"'"),l.assert("NHWC"===s,"Error in conv2d: got dataFormat of "+s+" but only NHWC is currently supported.");var g=c.computeConv2DInfo(d.shape,n.shape,r,p,o,h),m=u.ENV.engine.runKernel(function(e){return e.conv2d(d,n,g)},{x:d,filter:n},function(t){return l.assert(i(p),"Error in gradient of conv2D: dilation rates greater than 1 are notyet supported in gradients. Got dilations '"+p+"'"),{x:function(){return e.conv2dDerInput(d.shape,t,n,r,o)},filter:function(){return e.conv2dDerFilter(d,t,n.shape,r,o)}}});return f?m.as3D(m.shape[1],m.shape[2],m.shape[3]):m},e.conv2dDerInput=function(e,t,n,r,i,a){l.assertArgumentsAreTensors({dy:t,filter:n},"conv2dDerInput"),l.assert(e.length===t.rank,"Length of inShape ("+e.length+") and rank of dy ("+t.rank+") must match");var o=e,s=t,p=!1;3===t.rank&&(p=!0,s=t.as4D(1,t.shape[0],t.shape[1],t.shape[2]),o=[1,e[0],e[1],e[2]]);var h=o[3],d=s.shape[3];l.assert(4===o.length,"Error in conv2dDerInput: inShape must be length 4, but got length "+o.length+"."),l.assert(4===s.rank,"Error in conv2dDerInput: dy must be rank 4, but got rank "+s.rank),l.assert(4===n.rank,"Error in conv2dDerInput: filter must be rank 4, but got rank "+n.rank),l.assert(h===n.shape[2],"Error in conv2dDerInput: depth of input ("+h+") must match input depth for filter "+n.shape[2]+"."),l.assert(d===n.shape[3],"Error in conv2dDerInput: depth of output ("+d+") mustmatch output depth for filter "+n.shape[3]+"."),null!=a&&l.assert(l.isInt(i),"Error in conv2dDerInput: pad must be an integer when using, dimRoundingMode "+a+" but got pad "+i+".");var f=c.computeConv2DInfo(o,n.shape,r,1,i,a),g=u.ENV.engine.runKernel(function(e){return e.conv2dDerInput(s,n,f)},{dy4D:s});return p?g.as3D(g.shape[1],g.shape[2],g.shape[3]):g},e.conv2dDerFilter=function(e,t,n,r,i,a){l.assertArgumentsAreTensors({x:e,dy:t},"conv2dDerFilter");var o=e;3===e.rank&&(o=e.as4D(1,e.shape[0],e.shape[1],e.shape[2]));var s=t;3===s.rank&&(s=t.as4D(1,t.shape[0],t.shape[1],t.shape[2])),l.assert(4===o.rank,"Error in conv2dDerFilter: input must be rank 4, but got shape "+o.shape+"."),l.assert(4===s.rank,"Error in conv2dDerFilter: dy must be rank 4, but got shape "+s.shape+"."),l.assert(4===n.length,"Error in conv2dDerFilter: filterShape must be length 4, but got "+n+"."),l.assert(o.shape[3]===n[2],"Error in conv2dDerFilter: depth of input "+o.shape[3]+") must match input depth in filter ("+n[2]+"."),l.assert(s.shape[3]===n[3],"Error in conv2dDerFilter: depth of dy ("+s.shape[3]+") must match output depth for filter ("+n[3]+")."),null!=a&&l.assert(l.isInt(i),"Error in conv2dDerFilter: pad must be an integer when using, dimRoundingMode "+a+" but got pad "+i+".");var p=c.computeConv2DInfo(o.shape,n,r,1,i,a);return u.ENV.engine.runKernel(function(e){return e.conv2dDerFilter(o,s,p)},{x4D:o,dy4D:s})},e.conv2dTranspose=function(t,n,r,i,a,o){return l.assertArgumentsAreTensors({x:t,filter:n},"conv2dTranspose"),e.conv2dDerInput(r,t,n,i,a,o)},e.depthwiseConv2d=function(e,t,n,r,i,o,s){void 0===i&&(i="NHWC"),void 0===o&&(o=[1,1]),l.assertArgumentsAreTensors({x:e,filter:t},"depthwiseConv2d");var p=e,h=!1;3===e.rank&&(h=!0,p=e.as4D(1,e.shape[0],e.shape[1],e.shape[2])),l.assert(4===p.rank,"Error in depthwiseConv2D: input must be rank 4, but got rank "+p.rank+"."),l.assert(4===t.rank,"Error in depthwiseConv2D: filter must be rank 4, but got rank "+t.rank+"."),l.assert(p.shape[3]===t.shape[2],"Error in depthwiseConv2D: number of input channels ("+p.shape[3]+") must match the inChannels dimension in filter "+t.shape[2]+"."),null==o&&(o=[1,1]),l.assert(a(n,o),"Error in depthwiseConv2d: Either strides or dilations must be 1.Got strides "+n+" and dilations '"+o+"'"),null!=s&&l.assert(l.isInt(r),"Error in depthwiseConv2D: pad must be an integer when using, dimRoundingMode "+s+" but got pad "+r+".");var d=c.computeConv2DInfo(p.shape,t.shape,n,o,r,s,!0),f=u.ENV.engine.runKernel(function(e){return e.depthwiseConv2D(p,t,d)},{x4D:p,filter:t});return h?f.as3D(f.shape[1],f.shape[2],f.shape[3]):f},e.separableConv2d=function(t,n,r,i,a,o,s){void 0===o&&(o=[1,1]),void 0===s&&(s="NHWC"),l.assertArgumentsAreTensors({x:t,depthwiseFilter:n,pointwiseFilter:r},"separableConv2d");var u=t,c=!1;if(3===t.rank&&(c=!0,u=t.as4D(1,t.shape[0],t.shape[1],t.shape[2])),"NCHW"===s)throw new Error("separableConv2d currently does not support dataFormat NCHW; only NHWC is supported");l.assert(4===u.rank,"Error in separableConv2d: input must be rank 4, but got rank "+u.rank+"."),l.assert(4===n.rank,"Error in separableConv2d: depthwise filter must be rank 4, but got rank "+n.rank+"."),l.assert(4===r.rank,"Error in separableConv2d: pointwise filter must be rank 4, but got rank "+n.rank+"."),l.assert(1===r.shape[0],"Error in separableConv2d: the first dimension of pointwise filter  must be 1, but got "+r.shape[0]+"."),l.assert(1===r.shape[1],"Error in separableConv2d: the second dimension of pointwise filter  must be 1, but got "+r.shape[1]+".");var p=n.shape[2],h=n.shape[3];l.assert(r.shape[2]===p*h,"Error in separableConv2d: the third dimension of pointwise filter must be "+p*h+", but got "+r.shape[2]+".");var d=e.depthwiseConv2d(u,n,i,a,s,o),f=e.conv2d(d,r,1,"valid",s);return c?f.as3D(f.shape[1],f.shape[2],f.shape[3]):f},o([s.doc({heading:"Operations",subheading:"Convolution"}),p.operation],e,"conv1d",null),o([s.doc({heading:"Operations",subheading:"Convolution"}),p.operation],e,"conv2d",null),o([p.operation],e,"conv2dDerInput",null),o([p.operation],e,"conv2dDerFilter",null),o([s.doc({heading:"Operations",subheading:"Convolution"}),p.operation],e,"conv2dTranspose",null),o([s.doc({heading:"Operations",subheading:"Convolution"}),p.operation],e,"depthwiseConv2d",null),o([s.doc({heading:"Operations",subheading:"Convolution"}),p.operation],e,"separableConv2d",null),e}();n.ConvOps=h},{"../doc":3,"../environment":5,"../util":95,"./conv_util":55,"./operation":65}],55:[function(e,t,n){"use strict";function r(e,t,n,r,i,a,l,c){void 0===l&&(l=!1),void 0===c&&(c="channelsLast");var p=[-1,-1,-1,-1],h=p[0],d=p[1],f=p[2],g=p[3];if("channelsLast"===c)h=e[0],d=e[1],f=e[2],g=e[3];else{if("channelsFirst"!==c)throw new Error("Unknown dataFormat "+c);h=e[0],g=e[1],d=e[2],f=e[3]}var m,v=t[0],y=t[1],b=t[3],w=o(n),x=w[0],A=w[1],E=o(r),_=E[0],T=E[1],S=u(i,d,f,x,A,s(v,_),s(y,T),a),C=S.padInfo,O=S.outHeight,k=S.outWidth,R=l?b*g:b;return"channelsFirst"===c?m=[h,R,O,k]:"channelsLast"===c&&(m=[h,O,k,R]),{batchSize:h,dataFormat:c,inHeight:d,inWidth:f,inChannels:g,outHeight:O,outWidth:k,outChannels:R,padInfo:C,strideHeight:x,strideWidth:A,filterHeight:v,filterWidth:y,dilationHeight:_,dilationWidth:T,inShape:e,outShape:m,filterShape:t}}function i(e,t,n,r,i,o){null==i&&(i=a(e,t,r));var s=e[0],u=e[1],p=l((s-t+2*i)/r+1,o);c.assert(c.isInt(p),"The output # of rows ("+p+") must be an integer. Change the stride and/or zero pad parameters");var h=l((u-t+2*i)/r+1,o);return c.assert(c.isInt(h),"The output # of columns ("+h+") must be an integer. Change the stride and/or zero pad parameters"),[p,h,n]}function a(e,t,n,r){void 0===r&&(r=1);var i=s(t,r);return Math.floor((e[0]*(n-1)-n+i)/2)}function o(e){return"number"==typeof e?[e,e]:e}function s(e,t){return t<=1?e:e+(e-1)*(t-1)}function u(e,t,n,r,a,o,s,u){var l,c,p;if("number"==typeof e){l={top:e,bottom:e,left:e,right:e,type:0===e?"VALID":"NUMBER"};var h=i([t,n,1],o,1,r,e,u);c=h[0],p=h[1]}else if("same"===e){var d=((c=Math.ceil(t/r))-1)*r+o-t,f=((p=Math.ceil(n/a))-1)*a+s-n,g=Math.floor(d/2),m=d-g,v=Math.floor(f/2);l={top:g,bottom:m,left:v,right:f-v,type:"SAME"}}else{if("valid"!==e)throw Error("Unknown padding parameter: "+e);l={top:0,bottom:0,left:0,right:0,type:"VALID"},c=Math.ceil((t-o+1)/r),p=Math.ceil((n-s+1)/a)}return{padInfo:l,outHeight:c,outWidth:p}}function l(e,t){if(!t)return e;switch(t){case"round":return Math.round(e);case"ceil":return Math.ceil(e);case"floor":return Math.floor(e);default:throw new Error("Unknown roundingMode "+t)}}Object.defineProperty(n,"__esModule",{value:!0});var c=e("../util");n.computePool2DInfo=function(e,t,n,i,a,s){void 0===s&&(s="channelsLast");var u,l=o(t),c=l[0],p=l[1];if("channelsLast"===s)u=[c,p,e[3],e[3]];else{if("channelsFirst"!==s)throw new Error("Unknown dataFormat "+s);u=[c,p,e[1],e[1]]}return r(e,u,n,1,i,a,!1,s)},n.computeConv2DInfo=r,n.computeDefaultPad=a},{"../util":95}],56:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0}),n.ERF_P=.3275911,n.ERF_A1=.254829592,n.ERF_A2=-.284496736,n.ERF_A3=1.421413741,n.ERF_A4=-1.453152027,n.ERF_A5=1.061405429},{}],57:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./operation"),u=function(){function e(){}return e.resizeBilinear=function(e,t,n){void 0===n&&(n=!1),o.assertArgumentsAreTensors({images:e},"resizeBilinear"),o.assert(3===e.rank||4===e.rank,"Error in resizeBilinear: x must be rank 3 or 4, but got rank "+e.rank+"."),o.assert(2===t.length,"Error in resizeBilinear: new shape must 2D, but got shape "+t+".");var r=e,i=!1;3===e.rank&&(i=!0,r=e.as4D(1,e.shape[0],e.shape[1],e.shape[2]));var s=t[0],u=t[1],l=a.ENV.engine.runKernel(function(e){return e.resizeBilinear(r,s,u,n)},{batchImages:r});return i?l.as3D(l.shape[1],l.shape[2],l.shape[3]):l},e.resizeNearestNeighbor=function(e,t,n){void 0===n&&(n=!1),o.assertArgumentsAreTensors({images:e},"resizeNearestNeighbor"),o.assert(3===e.rank||4===e.rank,"Error in resizeNearestNeighbor: x must be rank 3 or 4, but got rank "+e.rank+"."),o.assert(2===t.length,"Error in resizeNearestNeighbor: new shape must 2D, but got shape "+t+"."),o.assert("float32"===e.dtype||"int32"===e.dtype,"`images` must have `int32` or `float32` as dtype");var r=e,i=!1;3===e.rank&&(i=!0,r=e.as4D(1,e.shape[0],e.shape[1],e.shape[2]));var s=t[0],u=t[1],l=a.ENV.engine.runKernel(function(e){return e.resizeNearestNeighbor(r,s,u,n)},{batchImages:r});return i?l.as3D(l.shape[1],l.shape[2],l.shape[3]):l},r([i.doc({heading:"Operations",subheading:"Images",namespace:"image"}),s.operation],e,"resizeBilinear",null),r([i.doc({heading:"Operations",subheading:"Images",namespace:"image"}),s.operation],e,"resizeNearestNeighbor",null),e}();n.ImageOps=u},{"../doc":3,"../environment":5,"../util":95,"./operation":65}],58:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../types"),s=e("../util"),u=e("./broadcast_util"),l=e("./operation"),c=function(){function e(){}return e.logicalNot=function(e){return s.assertArgumentsAreTensors({x:e},"logicalNot"),s.assert("bool"===e.dtype,"Error Array must be of type bool."),a.ENV.engine.runKernel(function(t){return t.logicalNot(e)},{x:e})},e.logicalAnd=function(e,t){return s.assertArgumentsAreTensors({a:e,b:t},"logicalAnd"),s.assert("bool"===e.dtype&&"bool"===t.dtype,"Error Array must be of type bool."),u.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.logicalAnd(e,t)},{a:e,b:t})},e.logicalOr=function(e,t){return s.assertArgumentsAreTensors({a:e,b:t},"logicalOr"),s.assert("bool"===e.dtype&&"bool"===t.dtype,"Error Array must be of type bool."),u.assertAndGetBroadcastShape(e.shape,t.shape),a.ENV.engine.runKernel(function(n){return n.logicalOr(e,t)},{a:e,b:t})},e.logicalXor=function(t,n){return s.assertArgumentsAreTensors({a:t,b:n},"logicalXor"),s.assert("bool"===t.dtype&&"bool"===n.dtype,"Error Array must be of type bool."),u.assertAndGetBroadcastShape(t.shape,n.shape),e.logicalOr(t,n).logicalAnd(e.logicalAnd(t,n).logicalNot())},e.where=function(e,t,n){s.assertArgumentsAreTensors({condition:e,a:t,b:n},"where"),s.assert("bool"===e.dtype||"bool"===t.dtype||"bool"===n.dtype,"Error Array must be of type bool."),s.assertShapesMatch(t.shape,n.shape,"Error in where: "),1===e.rank?s.assert(e.shape[0]===t.shape[0],"The first dimension of `a` must match the size of `condition`."):s.assertShapesMatch(e.shape,n.shape,"Error in where: ");var r=o.upcastType(t.dtype,n.dtype);return a.ENV.engine.runKernel(function(i){return i.where(e,t,n,r)},{condition:e,a:t,b:n})},r([i.doc({heading:"Operations",subheading:"Logical"}),l.operation],e,"logicalNot",null),r([i.doc({heading:"Operations",subheading:"Logical"}),l.operation],e,"logicalAnd",null),r([i.doc({heading:"Operations",subheading:"Logical"}),l.operation],e,"logicalOr",null),r([i.doc({heading:"Operations",subheading:"Logical"}),l.operation],e,"logicalXor",null),r([i.doc({heading:"Operations",subheading:"Logical"}),l.operation],e,"where",null),e}();n.LogicalOps=c},{"../doc":3,"../environment":5,"../types":94,"../util":95,"./broadcast_util":50,"./operation":65}],59:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i,a=e("../doc"),o=e("../util"),s=e("./operation"),u=e("./ops");!function(e){e[e.NONE=0]="NONE",e[e.MEAN=1]="MEAN",e[e.SUM=2]="SUM",e[e.SUM_BY_NONZERO_WEIGHTS=3]="SUM_BY_NONZERO_WEIGHTS"}(i=n.Reduction||(n.Reduction={}));var l=function(){function e(){}return e.computeWeightedLoss=function(e,t,n){void 0===n&&(n=i.SUM_BY_NONZERO_WEIGHTS),o.assertArgumentsAreTensors({losses:e},"computeWeightedLoss"),null!=t&&o.assertArgumentsAreTensors({weights:t},"computeWeightedLoss");var r=null==t?e:e.mul(t);if(n===i.NONE)return r;if(n===i.SUM)return r.sum();if(n===i.MEAN)return null==t?r.mean():r.sum().div(t.sum());if(n===i.SUM_BY_NONZERO_WEIGHTS){if(null==t)return r.sum().div(u.scalar(e.size));var a=t.notEqual(u.scalar(0)).sum().toFloat();return r.sum().div(a)}throw Error("Unknown reduction: "+n)},e.absoluteDifference=function(t,n,r,a){void 0===a&&(a=i.SUM_BY_NONZERO_WEIGHTS),o.assertArgumentsAreTensors({labels:t,predictions:n},"absoluteDifference"),null!=r&&o.assertArgumentsAreTensors({weights:r},"absoluteDifference"),o.assertShapesMatch(t.shape,n.shape,"Error in absoluteDifference: ");var s=t.sub(n).abs();return e.computeWeightedLoss(s,r,a)},r([a.doc({heading:"Training",subheading:"Losses",namespace:"losses"}),s.operation],e,"computeWeightedLoss",null),r([a.doc({heading:"Training",subheading:"Losses",namespace:"losses"}),s.operation],e,"absoluteDifference",null),e}();n.LossOps=l},{"../doc":3,"../util":95,"./operation":65,"./ops":66}],60:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./operation"),u=function(){function e(){}return e.localResponseNormalization=function(e,t,n,r,i){void 0===t&&(t=5),void 0===n&&(n=1),void 0===r&&(r=1),void 0===i&&(i=.5),o.assertArgumentsAreTensors({x:e},"localResponseNormalization"),o.assert(4===e.rank||3===e.rank,"Error in localResponseNormalization: x must be rank 3 or 4 but got\n               rank "+e.rank+"."),o.assert(o.isInt(t),"Error in localResponseNormalization3D: radius must be an integer\n                     but got radius "+t+".");var s=e,u=!1;3===e.rank&&(u=!0,s=e.as4D(1,e.shape[0],e.shape[1],e.shape[2]));var l=a.ENV.engine.runKernel(function(e){return e.localResponseNormalization4D(s,t,n,r,i)},{x4D:s});return u?l.as3D(l.shape[1],l.shape[2],l.shape[3]):l},r([i.doc({heading:"Operations",subheading:"Normalization"}),s.operation],e,"localResponseNormalization",null),e}();n.LRNOps=u},{"../doc":3,"../environment":5,"../util":95,"./operation":65}],61:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../util"),o=e("./operation"),s=function(){function e(){}return e.multiRNNCell=function(e,t,n,r){a.assertArgumentsAreTensors({data:t,c:n,h:r},"multiRNNCell");for(var i=t,o=[],s=0;s<e.length;s++){var u=e[s](i,n[s],r[s]);o.push(u[0]),o.push(u[1]),i=u[1]}for(var l=[],c=[],s=0;s<o.length;s+=2)l.push(o[s]),c.push(o[s+1]);return[l,c]},e.basicLSTMCell=function(e,t,n,r,i,o){a.assertArgumentsAreTensors({forgetBias:e,lstmKernel:t,lstmBias:n,data:r,c:i,h:o},"basicLSTMCell");var s=r.concat(o,1).matMul(t).add(n),u=s.shape[0],l=s.shape[1]/4,c=[u,l],p=s.slice([0,0],c),h=s.slice([0,l],c),d=s.slice([0,2*l],c),f=s.slice([0,3*l],c),g=p.sigmoid().mulStrict(h.tanh()).addStrict(i.mulStrict(e.add(d).sigmoid()));return[g,g.tanh().mulStrict(f.sigmoid())]},r([i.doc({heading:"Operations",subheading:"RNN"}),o.operation],e,"multiRNNCell",null),r([i.doc({heading:"Operations",subheading:"RNN"}),o.operation],e,"basicLSTMCell",null),e}();n.LSTMOps=s},{"../doc":3,"../util":95,"./operation":65}],62:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./operation"),u=function(){function e(){}return e.matMul=function(e,t,n,r){void 0===n&&(n=!1),void 0===r&&(r=!1),o.assertArgumentsAreTensors({a:e,b:t},"matMul");var i=n?e.shape[0]:e.shape[1],s=r?t.shape[1]:t.shape[0];o.assert(2===e.rank&&2===t.rank,"Error in matMul: inputs must be rank 2, got ranks "+e.rank+" and "+t.rank+"."),o.assert(i===s,"Error in matMul: inner shapes ("+i+") and ("+s+") of Tensors with shapes "+e.shape+" and "+t.shape+" and transposeA="+n+" and transposeB="+r+" must match.");return a.ENV.engine.runKernel(function(i){return i.matMul(e,t,n,r)},{a:e,b:t},function(i){return n||r?!n&&r?{a:function(){return i.matMul(t.toFloat(),!1,!1)},b:function(){return i.matMul(e.toFloat(),!0,!1)}}:n&&!r?{a:function(){return t.toFloat().matMul(i,!1,!0)},b:function(){return e.toFloat().matMul(i,!1,!1)}}:{a:function(){return t.toFloat().matMul(i,!0,!0)},b:function(){return i.matMul(e.toFloat(),!0,!0)}}:{a:function(){return i.matMul(t.toFloat(),!1,!0)},b:function(){return e.toFloat().matMul(i,!0,!1)}}})},e.vectorTimesMatrix=function(e,t){return o.assert(1===e.rank,"Error in vectorTimesMatrix: first input must be rank 1, but got rank "+e.rank+"."),o.assert(2===t.rank,"Error in vectorTimesMatrix: second input must be rank 2, but got rank "+t.rank+"."),o.assert(e.size===t.shape[0],"Error in vectorTimesMatrix: size of vector ("+e.size+") must match first dimension of matrix ("+t.shape[0]+")"),e.as2D(1,-1).matMul(t).as1D()},e.matrixTimesVector=function(e,t){return o.assert(1===t.rank,"Error in matrixTimesVector: second input must rank 1, but got rank "+t.rank+"."),o.assert(2===e.rank,"Error in matrixTimesVector: first input must be a rank 2, but got rank "+e.rank+"."),o.assert(t.size===e.shape[1],"Error in matrixTimesVector: size of first rank 1 input "+t.size+" must match inner dimension of second rank 2 input, but got shape "+e.shape+"."),e.matMul(t.as2D(-1,1)).as1D()},e.dotProduct=function(e,t){return o.assert(1===e.rank&&1===t.rank,"Error in dotProduct: inputs must be rank 1, but got ranks "+e.rank+" and "+t.rank+"."),o.assert(e.size===t.size,"Error in dotProduct: size of inputs ("+e.size+") and ("+t.size+") must match."),e.as2D(1,-1).matMul(t.as2D(-1,1)).asScalar()},e.outerProduct=function(e,t){return o.assert(1===e.rank&&1===t.rank,"Error in outerProduct: inputs must be rank 1, but got ranks "+e.rank+" and "+t.rank+"."),e.as2D(-1,1).matMul(t.as2D(1,-1))},r([i.doc({heading:"Operations",subheading:"Matrices"}),s.operation],e,"matMul",null),r([s.operation],e,"vectorTimesMatrix",null),r([s.operation],e,"matrixTimesVector",null),r([s.operation],e,"dotProduct",null),r([i.doc({heading:"Operations",subheading:"Matrices"}),s.operation],e,"outerProduct",null),e}();n.MatmulOps=u},{"../doc":3,"../environment":5,"../util":95,"./operation":65}],63:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../util"),o=e("./array_ops"),s=e("./binary_ops"),u=e("./operation"),l=function(){function e(){}return e.movingAverage=function(e,t,n,r,i){void 0===i&&(i=!0),a.assertArgumentsAreTensors({v:e,x:t},"movingAverage"),a.assertTypesMatch(e,t),a.assert(a.arraysEqual(e.shape,t.shape),"Shape mismatch in v and x");var u=o.ArrayOps.scalar(1);n="number"==typeof n?o.ArrayOps.scalar(n):n;var l=u.sub(n),c=t.sub(e).mul(l);return i&&(a.assert(null!=r,"When using zeroDebias: true, step is required."),r="number"==typeof r?o.ArrayOps.scalar(r):r,c=c.div(u.sub(s.BinaryOps.pow(n,r)))),e.add(c)},r([i.doc({heading:"Operations",subheading:"Moving Average"}),u.operation],e,"movingAverage",null),e}();n.MovingAverageOps=l},{"../doc":3,"../util":95,"./array_ops":46,"./binary_ops":49,"./operation":65}],64:[function(e,t,n){"use strict";function r(e,t,n){if(void 0===n&&(n=null),0===e.rank)return e.abs();if(1!==e.rank&&null===n)return r(e.reshape([-1]),t,n);if(1===e.rank||"number"==typeof n||n instanceof Array&&1===n.length){if(1===t)return e.abs().sum(n);if(t===1/0)return e.abs().max(n);if(t===-1/0)return e.abs().min(n);if("euclidean"===t||2===t)return e.abs().pow(l.scalar(2,"int32")).sum(n).sqrt();throw new Error("Error in norm: invalid ord value: "+t)}if(n instanceof Array&&2===n.length){if(1===t)return e.abs().sum(n[0]).max(n[1]-1);if(t===1/0)return e.abs().sum(n[1]).max(n[0]);if(t===-1/0)return e.abs().sum(n[1]).min(n[0]);if("fro"===t||"euclidean"===t)return e.square().sum(n).sqrt();throw new Error("Error in norm: invalid ord value: "+t)}throw new Error("Error in norm: invalid axis: "+n)}var i=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var a=e("../doc"),o=e("../util"),s=e("./axis_util"),u=e("./operation"),l=e("./ops"),c=function(){function e(){}return e.norm=function(e,t,n,i){void 0===t&&(t="euclidean"),void 0===n&&(n=null),void 0===i&&(i=!1),o.assertArgumentsAreTensors({x:e},"norm");var a=r(e,t,n),u=a.shape;if(i){var l=s.parseAxisParam(n,e.shape);u=s.expandShapeToKeepDim(a.shape,l)}return a.reshape(u)},i([a.doc({heading:"Operations",subheading:"Matrices"}),u.operation],e,"norm",null),e}();n.NormOps=c},{"../doc":3,"../util":95,"./axis_util":47,"./operation":65,"./ops":66}],65:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../globals");n.operation=function(e,t,n){var i=n.value;return n.value=function(){for(var e=[],n=0;n<arguments.length;n++)e[n]=arguments[n];return r.tidy(t,function(){return i.apply(void 0,e)})},n}},{"../globals":6}],66:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./array_ops"),i=e("./batchnorm"),a=e("./binary_ops"),o=e("./compare"),s=e("./concat"),u=e("./conv"),l=e("./image_ops"),c=e("./logical_ops"),p=e("./loss_ops"),h=e("./lrn"),d=e("./lstm"),f=e("./matmul"),g=e("./moving_average"),m=e("./norm"),v=e("./pool"),y=e("./reduction_ops"),b=e("./reverse"),w=e("./slice"),x=e("./softmax"),A=e("./transpose"),E=e("./unary_ops");n.batchNormalization=i.BatchNormOps.batchNormalization,n.batchNormalization2d=i.BatchNormOps.batchNormalization2d,n.batchNormalization3d=i.BatchNormOps.batchNormalization3d,n.batchNormalization4d=i.BatchNormOps.batchNormalization4d,n.concat=s.ConcatOps.concat,n.concat1d=s.ConcatOps.concat1d,n.concat2d=s.ConcatOps.concat2d,n.concat3d=s.ConcatOps.concat3d,n.concat4d=s.ConcatOps.concat4d,n.conv1d=u.ConvOps.conv1d,n.conv2d=u.ConvOps.conv2d,n.conv2dTranspose=u.ConvOps.conv2dTranspose,n.depthwiseConv2d=u.ConvOps.depthwiseConv2d,n.separableConv2d=u.ConvOps.separableConv2d,n.matMul=f.MatmulOps.matMul,n.matrixTimesVector=f.MatmulOps.matrixTimesVector,n.outerProduct=f.MatmulOps.outerProduct,n.vectorTimesMatrix=f.MatmulOps.vectorTimesMatrix,n.avgPool=v.PoolOps.avgPool,n.maxPool=v.PoolOps.maxPool,n.transpose=A.TransposeOps.transpose,n.reverse=b.ReverseOps.reverse,n.reverse1d=b.ReverseOps.reverse1d,n.reverse2d=b.ReverseOps.reverse2d,n.reverse3d=b.ReverseOps.reverse3d,n.reverse4d=b.ReverseOps.reverse4d,n.slice=w.SliceOps.slice,n.slice1d=w.SliceOps.slice1d,n.slice2d=w.SliceOps.slice2d,n.slice3d=w.SliceOps.slice3d,n.slice4d=w.SliceOps.slice4d,n.argMax=y.ReductionOps.argMax,n.argMin=y.ReductionOps.argMin,n.logSumExp=y.ReductionOps.logSumExp,n.max=y.ReductionOps.max,n.mean=y.ReductionOps.mean,n.min=y.ReductionOps.min,n.moments=y.ReductionOps.moments,n.sum=y.ReductionOps.sum,n.equal=o.CompareOps.equal,n.equalStrict=o.CompareOps.equalStrict,n.greater=o.CompareOps.greater,n.greaterStrict=o.CompareOps.greaterStrict,n.greaterEqual=o.CompareOps.greaterEqual,n.greaterEqualStrict=o.CompareOps.greaterEqualStrict,n.less=o.CompareOps.less,n.lessStrict=o.CompareOps.lessStrict,n.lessEqual=o.CompareOps.lessEqual,n.lessEqualStrict=o.CompareOps.lessEqualStrict,n.notEqual=o.CompareOps.notEqual,n.notEqualStrict=o.CompareOps.notEqualStrict,n.logicalNot=c.LogicalOps.logicalNot,n.logicalAnd=c.LogicalOps.logicalAnd,n.logicalOr=c.LogicalOps.logicalOr,n.logicalXor=c.LogicalOps.logicalXor,n.where=c.LogicalOps.where,n.abs=E.UnaryOps.abs,n.acos=E.UnaryOps.acos,n.acosh=E.UnaryOps.acosh,n.asin=E.UnaryOps.asin,n.asinh=E.UnaryOps.asinh,n.atan=E.UnaryOps.atan,n.atanh=E.UnaryOps.atanh,n.ceil=E.UnaryOps.ceil,n.clipByValue=E.UnaryOps.clipByValue,n.cos=E.UnaryOps.cos,n.cosh=E.UnaryOps.cosh,n.elu=E.UnaryOps.elu,n.exp=E.UnaryOps.exp,n.expm1=E.UnaryOps.expm1,n.floor=E.UnaryOps.floor,n.sign=E.UnaryOps.sign,n.leakyRelu=E.UnaryOps.leakyRelu,n.log=E.UnaryOps.log,n.log1p=E.UnaryOps.log1p,n.logSigmoid=E.UnaryOps.logSigmoid,n.neg=E.UnaryOps.neg,n.prelu=E.UnaryOps.prelu,n.relu=E.UnaryOps.relu,n.reciprocal=E.UnaryOps.reciprocal,n.round=E.UnaryOps.round,n.selu=E.UnaryOps.selu,n.sigmoid=E.UnaryOps.sigmoid,n.sin=E.UnaryOps.sin,n.sinh=E.UnaryOps.sinh,n.softplus=E.UnaryOps.softplus,n.sqrt=E.UnaryOps.sqrt,n.rsqrt=E.UnaryOps.rsqrt,n.square=E.UnaryOps.square,n.step=E.UnaryOps.step,n.tan=E.UnaryOps.tan,n.tanh=E.UnaryOps.tanh,n.erf=E.UnaryOps.erf,n.add=a.BinaryOps.add,n.addStrict=a.BinaryOps.addStrict,n.atan2=a.BinaryOps.atan2,n.div=a.BinaryOps.div,n.divStrict=a.BinaryOps.divStrict,n.maximum=a.BinaryOps.maximum,n.maximumStrict=a.BinaryOps.maximumStrict,n.minimum=a.BinaryOps.minimum,n.minimumStrict=a.BinaryOps.minimumStrict,n.mod=a.BinaryOps.mod,n.modStrict=a.BinaryOps.modStrict,n.mul=a.BinaryOps.mul,n.mulStrict=a.BinaryOps.mulStrict,n.pow=a.BinaryOps.pow,n.powStrict=a.BinaryOps.powStrict,n.sub=a.BinaryOps.sub,n.subStrict=a.BinaryOps.subStrict,n.squaredDifference=a.BinaryOps.squaredDifference,n.squaredDifferenceStrict=a.BinaryOps.squaredDifferenceStrict,n.norm=m.NormOps.norm,n.cast=r.ArrayOps.cast,n.clone=r.ArrayOps.clone,n.fromPixels=r.ArrayOps.fromPixels,n.toPixels=r.ArrayOps.toPixels,n.ones=r.ArrayOps.ones,n.onesLike=r.ArrayOps.onesLike,n.zeros=r.ArrayOps.zeros,n.zerosLike=r.ArrayOps.zerosLike,n.rand=r.ArrayOps.rand,n.randomNormal=r.ArrayOps.randomNormal,n.truncatedNormal=r.ArrayOps.truncatedNormal,n.randomUniform=r.ArrayOps.randomUniform,n.multinomial=r.ArrayOps.multinomial,n.reshape=r.ArrayOps.reshape,n.squeeze=r.ArrayOps.squeeze,n.tile=r.ArrayOps.tile,n.gather=r.ArrayOps.gather,n.oneHot=r.ArrayOps.oneHot,n.linspace=r.ArrayOps.linspace,n.range=r.ArrayOps.range,n.buffer=r.ArrayOps.buffer,n.fill=r.ArrayOps.fill,n.tensor=r.ArrayOps.tensor,n.scalar=r.ArrayOps.scalar,n.tensor1d=r.ArrayOps.tensor1d,n.tensor2d=r.ArrayOps.tensor2d,n.tensor3d=r.ArrayOps.tensor3d,n.tensor4d=r.ArrayOps.tensor4d,n.print=r.ArrayOps.print,n.expandDims=r.ArrayOps.expandDims,n.stack=r.ArrayOps.stack,n.split=r.ArrayOps.split,n.pad=r.ArrayOps.pad,n.pad1d=r.ArrayOps.pad1d,n.pad2d=r.ArrayOps.pad2d,n.pad3d=r.ArrayOps.pad3d,n.pad4d=r.ArrayOps.pad4d,n.movingAverage=g.MovingAverageOps.movingAverage,n.basicLSTMCell=d.LSTMOps.basicLSTMCell,n.multiRNNCell=d.LSTMOps.multiRNNCell,n.softmax=x.SoftmaxOps.softmax,n.localResponseNormalization=h.LRNOps.localResponseNormalization;var _=e("../tensor"),T=e("../types");_.Tensor,T.Rank,p.Reduction,n.losses={softmaxCrossEntropy:x.SoftmaxOps.softmaxCrossEntropy,absoluteDifference:p.LossOps.absoluteDifference,computeWeightedLoss:p.LossOps.computeWeightedLoss},n.image={resizeBilinear:l.ImageOps.resizeBilinear,resizeNearestNeighbor:l.ImageOps.resizeNearestNeighbor}},{"../tensor":89,"../types":94,"./array_ops":46,"./batchnorm":48,"./binary_ops":49,"./compare":51,"./concat":52,"./conv":54,"./image_ops":57,"./logical_ops":58,"./loss_ops":59,"./lrn":60,"./lstm":61,"./matmul":62,"./moving_average":63,"./norm":64,"./pool":67,"./reduction_ops":70,"./reverse":71,"./slice":73,"./softmax":75,"./transpose":76,"./unary_ops":77}],67:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./conv_util"),u=e("./operation"),l=function(){function e(){}return e.maxPool=function(t,n,r,i,u){o.assertArgumentsAreTensors({x:t},"maxPool");var l=t,c=!1;3===t.rank&&(c=!0,l=t.as4D(1,t.shape[0],t.shape[1],t.shape[2])),o.assert(4===l.rank,"Error in maxPool: input must be rank 4 but got rank "+l.rank+"."),null!=u&&o.assert(o.isInt(i),"Error in maxPool: pad must be an integer when using, dimRoundingMode "+u+" but got pad "+i+".");var p=s.computePool2DInfo(l.shape,n,r,i,u),h=a.ENV.engine.runKernel(function(e,t){return t(e.maxPool(l,p))},{x:l},function(t,a){var o=a[0];return{x:function(){return e.maxPoolBackprop(t,l,o,n,r,i)}}});return c?h.as3D(h.shape[1],h.shape[2],h.shape[3]):h},e.maxPoolBackprop=function(e,t,n,r,i,u,l){o.assertArgumentsAreTensors({dy:e,input:t,output:n},"maxPoolBackprop"),o.assert(t.rank===e.rank,"Rank of input ("+t.rank+") does not match rank of dy ("+e.rank+")"),o.assert(4===e.rank,"Error in maxPoolBackprop: dy must be rank 4 but got rank "+e.rank+"."),o.assert(4===t.rank,"Error in maxPoolBackprop: input must be rank 4 but got rank "+t.rank+"."),null!=l&&o.assert(o.isInt(u),"Error in maxPoolBackprop: pad must be an integer when using, dimRoundingMode "+l+" but got pad "+u+".");var c=s.computePool2DInfo(t.shape,r,i,u,l);return a.ENV.engine.runKernel(function(r){return r.maxPoolBackprop(e,t,n,c)},{dy:e,input:t})},e.avgPool=function(t,n,r,i,u){o.assertArgumentsAreTensors({x:t},"avgPool"),o.assert("float32"===t.dtype,"The input dtype to avgPool must be float32");var l=t,c=!1;3===t.rank&&(c=!0,l=t.as4D(1,t.shape[0],t.shape[1],t.shape[2])),o.assert(4===l.rank,"Error in avgPool: x must be rank 4 but got rank "+l.rank+"."),null!=u&&o.assert(o.isInt(i),"Error in avgPool: pad must be an integer when using, dimRoundingMode "+u+" but got pad "+i+".");var p=s.computePool2DInfo(l.shape,n,r,i),h=a.ENV.engine.runKernel(function(e){return e.avgPool(l,p)},{x:l},function(t){return{x:function(){return e.avgPoolBackprop(t,l,n,r,i)}}});return h=h.cast(t.dtype),c?h.as3D(h.shape[1],h.shape[2],h.shape[3]):h},e.avgPoolBackprop=function(e,t,n,r,i){o.assertArgumentsAreTensors({dy:e,input:t},"avgPoolBackprop"),o.assert(t.rank===e.rank,"Rank of input ("+t.rank+") does not match rank of dy ("+e.rank+")");var u=t,l=e,c=!1;3===t.rank&&(c=!0,u=t.as4D(1,t.shape[0],t.shape[1],t.shape[2]),l=e.as4D(1,e.shape[0],e.shape[1],e.shape[2])),o.assert(4===l.rank,"Error in avgPoolBackprop: dy must be rank 4 but got rank "+l.rank+"."),o.assert(4===u.rank,"Error in avgPoolBackprop: input must be rank 4 but got rank "+u.rank+".");var p=s.computePool2DInfo(u.shape,n,r,i),h=a.ENV.engine.runKernel(function(e){return e.avgPoolBackprop(l,u,p)},{dy4D:l,input4D:u});return c?h.as3D(h.shape[1],h.shape[2],h.shape[3]):h},r([i.doc({heading:"Operations",subheading:"Convolution"}),u.operation],e,"maxPool",null),r([u.operation],e,"maxPoolBackprop",null),r([i.doc({heading:"Operations",subheading:"Convolution"}),u.operation],e,"avgPool",null),r([u.operation],e,"avgPoolBackprop",null),e}();n.PoolOps=l},{"../doc":3,"../environment":5,"../util":95,"./conv_util":55,"./operation":65}],68:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("seedrandom"),i=function(){function e(e,t,n,i,a){this.mean=e,this.stdDev=t,this.dtype=n,this.nextVal=NaN,this.truncated=i,this.truncated&&(this.upper=this.mean+2*this.stdDev,this.lower=this.mean-2*this.stdDev);var o=a||Math.random();this.random=r.alea(o.toString())}return e.prototype.nextValue=function(){if(!isNaN(this.nextVal)){var e=this.nextVal;return this.nextVal=NaN,e}for(var t,n,r=!1;!r;){var i=void 0,a=void 0,o=void 0;do{o=(i=2*this.random()-1)*i+(a=2*this.random()-1)*a}while(o>=1||0===o);var s=Math.sqrt(-2*Math.log(o)/o);t=this.mean+this.stdDev*i*s,n=this.mean+this.stdDev*a*s,this.truncated&&!this.isValidTruncated(t)||(r=!0)}return this.truncated&&!this.isValidTruncated(n)||(this.nextVal=this.convertValue(n)),this.convertValue(t)},e.prototype.convertValue=function(e){return null==this.dtype||"float32"===this.dtype?e:Math.round(e)},e.prototype.isValidTruncated=function(e){return e<=this.upper&&e>=this.lower},e}();n.MPRandGauss=i},{seedrandom:135}],69:[function(e,t,n){"use strict";function r(e,t){for(var n=t;n<e;++n)if(e%n==0)return n;return e}Object.defineProperty(n,"__esModule",{value:!0}),n.PARALLELIZE_THRESHOLD=30,n.computeOptimalWindowSize=function(e){return e<=n.PARALLELIZE_THRESHOLD?e:r(e,Math.floor(Math.sqrt(e)))}},{}],70:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../globals"),s=e("../util"),u=e("./axis_util"),l=e("./operation"),c=e("./ops"),p=function(){function e(){}return e.logSumExp=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"logSumExp");var r=u.parseAxisParam(t,e.shape),i=e.max(r,!0),a=e.sub(i).exp().sum(r).log(),o=i.reshape(a.shape).add(a);if(n){var l=u.expandShapeToKeepDim(o.shape,r);return o.reshape(l)}return o},e.sum=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"sum"),"bool"===e.dtype&&(e=e.toInt());var r=u.parseAxisParam(t,e.shape);return o.customGrad(function(e){var t=u.getAxesPermutation(r,e.rank),i=r,o=e;null!=t&&(o=e.transpose(t),i=u.getInnerMostAxes(i.length,e.rank));var s=a.ENV.engine.runKernel(function(e){return e.sum(o,i)},{permutedX:o});if(n){var l=u.expandShapeToKeepDim(s.shape,r);s=s.reshape(l)}return{value:s,gradFunc:function(t){var n=e.shape.slice();return r.forEach(function(e){n[e]=1}),t.reshape(n).mul(c.ones(e.shape,"float32"))}}})(e)},e.mean=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"mean");var r=u.parseAxisParam(t,e.shape),i=u.computeOutAndReduceShapes(e.shape,r)[1],a=s.sizeFromShape(i);return o.customGrad(function(e){var i=c.scalar(a);return{value:(i.dtype===e.dtype?e:e.cast(i.dtype)).div(i).sum(t,n),gradFunc:function(t){var n=e.shape.slice();return r.forEach(function(e){n[e]=1}),t.reshape(n).mul(c.ones(e.shape,"float32")).div(i)}}})(e)},e.min=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"min");var r=u.parseAxisParam(t,e.shape),i=r,o=u.getAxesPermutation(i,e.rank);null!=o&&(e=e.transpose(o),i=u.getInnerMostAxes(i.length,e.rank));var l=a.ENV.engine.runKernel(function(t){return t.min(e,i)},{x:e});if(n){var c=u.expandShapeToKeepDim(l.shape,r);return l.reshape(c)}return l},e.max=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"max");var r=u.parseAxisParam(t,e.shape),i=r,o=u.getAxesPermutation(i,e.rank);null!=o&&(e=e.transpose(o),i=u.getInnerMostAxes(i.length,e.rank));var l=a.ENV.engine.runKernel(function(t){return t.max(e,i)},{x:e});if(n){var c=u.expandShapeToKeepDim(l.shape,r);return l.reshape(c)}return l},e.argMin=function(e,t){void 0===t&&(t=0),s.assertArgumentsAreTensors({x:e},"argMin"),null==t&&(t=0);var n=u.parseAxisParam(t,e.shape),r=u.getAxesPermutation(n,e.rank);return null!=r&&(e=e.transpose(r),n=u.getInnerMostAxes(n.length,e.rank)),a.ENV.engine.runKernel(function(t){return t.argMin(e,n[0])},{x:e})},e.argMax=function(e,t){void 0===t&&(t=0),s.assertArgumentsAreTensors({x:e},"argMax"),null==t&&(t=0);var n=u.parseAxisParam(t,e.shape),r=u.getAxesPermutation(n,e.rank);return null!=r&&(e=e.transpose(r),n=u.getInnerMostAxes(n.length,e.rank)),a.ENV.engine.runKernel(function(t){return t.argMax(e,n[0])},{x:e})},e.moments=function(e,t,n){void 0===t&&(t=null),void 0===n&&(n=!1),s.assertArgumentsAreTensors({x:e},"moments");var r=u.parseAxisParam(t,e.shape),i=e.mean(r,n),a=i.shape;return n||(a=u.expandShapeToKeepDim(i.shape,r)),{mean:i,variance:e.toFloat().sub(i.reshape(a)).square().mean(r,n)}},r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"logSumExp",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"sum",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"mean",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"min",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"max",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"argMin",null),r([i.doc({heading:"Operations",subheading:"Reduction"}),l.operation],e,"argMax",null),r([i.doc({heading:"Operations",subheading:"Normalization"}),l.operation],e,"moments",null),e}();n.ReductionOps=p},{"../doc":3,"../environment":5,"../globals":6,"../util":95,"./axis_util":47,"./operation":65,"./ops":66}],71:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./axis_util"),u=e("./operation"),l=function(){function e(){}return e.reverse1d=function(t){return o.assert(1===t.rank,"Error in reverse1D: x must be rank 1 but got\n             rank "+t.rank+"."),e.reverse(t,0)},e.reverse2d=function(t,n){return o.assert(2===t.rank,"Error in reverse2D: x must be rank 2 but got\n             rank "+t.rank+"."),e.reverse(t,n)},e.reverse3d=function(t,n){return o.assert(3===t.rank,"Error in reverse3D: x must be rank 3 but got\n             rank "+t.rank+"."),e.reverse(t,n)},e.reverse4d=function(t,n){return o.assert(4===t.rank,"Error in reverse4D: x must be rank 4 but got\n             rank "+t.rank+"."),e.reverse(t,n)},e.reverse=function(e,t){if(o.assertArgumentsAreTensors({x:e},"reverse"),0===e.rank)return e.clone();var n=s.parseAxisParam(t,e.shape);return a.ENV.engine.runKernel(function(t){return t.reverse(e,n)},{x:e},function(e){return{x:function(){return e.reverse(n)}}}).reshapeAs(e)},r([i.doc({heading:"Tensors",subheading:"Slicing and Joining"}),u.operation],e,"reverse",null),e}();n.ReverseOps=l},{"../doc":3,"../environment":5,"../util":95,"./axis_util":47,"./operation":65}],72:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0}),n.SELU_SCALEALPHA=1.7580993408473768,n.SELU_SCALE=1.0507009873554805},{}],73:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./operation"),u=e("./slice_util"),l=function(){function e(){}return e.slice1d=function(t,n,r){return o.assert(1===t.rank,"slice1d expects a rank-1 tensor, but got a rank-"+t.rank+" tensor"),e.slice(t,[n],[r])},e.slice2d=function(t,n,r){return o.assert(2===t.rank,"slice1d expects a rank-2 tensor, but got a rank-"+t.rank+" tensor"),e.slice(t,n,r)},e.slice3d=function(t,n,r){return o.assert(3===t.rank,"slice1d expects a rank-3 tensor, but got a rank-"+t.rank+" tensor"),e.slice(t,n,r)},e.slice4d=function(t,n,r){return o.assert(4===t.rank,"slice1d expects a rank-4 tensor, but got a rank-"+t.rank+" tensor"),e.slice(t,n,r)},e.slice=function(e,t,n){if(o.assertArgumentsAreTensors({x:e},"slice"),0===e.rank)throw new Error("Slicing scalar is not possible");var r;r="number"==typeof t?[t].concat(new Array(e.rank-1).fill(0)):t.length<e.rank?t.concat(new Array(e.rank-t.length).fill(0)):t;var i;i=null==n?new Array(e.rank).fill(-1):"number"==typeof n?[n].concat(new Array(e.rank-1).fill(-1)):n.length<e.rank?n.concat(new Array(e.rank-n.length).fill(-1)):n,i=i.map(function(t,n){return t>=0?t:(o.assert(-1===t,"Bad value in size"),e.shape[n]-r[n])}),u.assertParamsValid(e,r,i);var s=e.shape;return a.ENV.engine.runKernel(function(t){return t.slice(e,r,i)},{x:e},function(e){for(var t=[],n=0;n<e.rank;n++)t.push([r[n],s[n]-r[n]-i[n]]);return{x:function(){return e.pad(t)}}})},r([i.doc({heading:"Tensors",subheading:"Slicing and Joining"}),s.operation],e,"slice",null),e}();n.SliceOps=l},{"../doc":3,"../environment":5,"../util":95,"./operation":65,"./slice_util":74}],74:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../util");n.assertParamsValid=function(e,t,n){r.assert(e.rank===t.length,"Error in slice"+e.rank+"D: Length of begin "+t+" must match the rank of the array ("+e.rank+")."),r.assert(e.rank===n.length,"Error in slice"+e.rank+"D: Length of size "+n+" must match the rank of the array ("+e.rank+").");for(var i=0;i<e.rank;++i)r.assert(t[i]+n[i]<=e.shape[i],"Error in slice"+e.rank+"D: begin["+i+"] + size["+i+"] ("+(t[i]+n[i])+") would overflow input.shape["+i+"] ("+e.shape[i]+")")}},{"../util":95}],75:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../globals"),o=e("../util"),s=e("./axis_util"),u=e("./operation"),l=e("./ops"),c=function(){function e(){}return e.softmax=function(e,t){if(void 0===t&&(t=-1),o.assertArgumentsAreTensors({logits:e},"softmax"),-1===t&&(t=e.rank-1),t!==e.rank-1)throw Error("Softmax along a non-last dimension is not yet supported. Logits was rank "+e.rank+" and dim was "+t);return a.customGrad(function(e){var n=e.logSumExp([t],!0),r=e.toFloat().sub(n).exp();return{value:r,gradFunc:function(e){var n=e.mul(r);return n.sub(n.sum([t],!0).mul(r))}}})(e)},e.softmaxCrossEntropy=function(e,t,n){if(void 0===n&&(n=-1),o.assertArgumentsAreTensors({labels:e,logits:t},"softmaxCrossEntropy"),o.assertShapesMatch(e.shape,t.shape,"Error in softmaxCrossEntropy: "),-1===n&&(n=t.rank-1),n!==t.rank-1)throw Error("Softmax cross entropy along a non-last dimension is not yet supported. Labels / logits was rank "+t.rank+" and dim was "+n);return a.customGrad(function(e,t){var r=t.softmax(n);return{value:l.scalar(1e-5).add(r).log().mul(e).neg().sum([n]),gradFunc:function(t){var i=s.expandShapeToKeepDim(t.shape,[n]);return[t.reshape(i).mul(e.toFloat().sub(r)),t.reshape(i).mul(r.sub(e.toFloat()))]}}})(e,t)},r([i.doc({heading:"Operations",subheading:"Normalization"}),u.operation],e,"softmax",null),r([i.doc({heading:"Training",subheading:"Losses",namespace:"losses"}),u.operation],e,"softmaxCrossEntropy",null),e}();n.SoftmaxOps=c},{"../doc":3,"../globals":6,"../util":95,"./axis_util":47,"./operation":65,"./ops":66}],76:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./axis_util"),u=e("./operation"),l=function(){function e(){}return e.transpose=function(e,t){o.assertArgumentsAreTensors({x:e},"transpose"),null==t&&(t=e.shape.map(function(e,t){return t}).reverse());return o.assert(e.rank===t.length,"Error in transpose: rank of input "+e.rank+" must match length of perm "+t+"."),a.ENV.engine.runKernel(function(n){return n.transpose(e,t)},{x:e},function(e){var n=s.getUndoAxesPermutation(t);return{x:function(){return e.transpose(n)}}})},r([i.doc({heading:"Operations",subheading:"Matrices"}),u.operation],e,"transpose",null),e}();n.TransposeOps=l},{"../doc":3,"../environment":5,"../util":95,"./axis_util":47,"./operation":65}],77:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../environment"),o=e("../util"),s=e("./operation"),u=e("./ops"),l=e("./ops"),c=e("./selu_util"),p=function(){function e(){}return e.neg=function(e){o.assertArgumentsAreTensors({x:e},"neg");return a.ENV.engine.runKernel(function(t){return t.neg(e)},{x:e},function(e){return{x:function(){return e.neg()}}})},e.ceil=function(e){o.assertArgumentsAreTensors({x:e},"ceil");return a.ENV.engine.runKernel(function(t){return t.ceil(e)},{x:e},function(e){return{x:function(){return u.zerosLike(e)}}})},e.floor=function(e){o.assertArgumentsAreTensors({x:e},"floor");return a.ENV.engine.runKernel(function(t){return t.floor(e)},{x:e},function(e){return{x:function(){return u.zerosLike(e)}}})},e.sign=function(e){o.assertArgumentsAreTensors({x:e},"sign");return a.ENV.engine.runKernel(function(t){return t.sign(e)},{x:e},function(e){return{x:function(){return u.zerosLike(e)}}})},e.round=function(e){o.assertArgumentsAreTensors({x:e},"round");return a.ENV.engine.runKernel(function(t){return t.round(e)},{x:e},function(e){return{x:function(){return u.zerosLike(e)}}})},e.exp=function(e){o.assertArgumentsAreTensors({x:e},"exp");return a.ENV.engine.runKernel(function(t,n){return n(t.exp(e))},{x:e},function(e,t){var n=t[0];return{x:function(){return e.mulStrict(n)}}})},e.expm1=function(e){o.assertArgumentsAreTensors({x:e},"expm1");return a.ENV.engine.runKernel(function(t){return t.expm1(e)},{x:e},function(t){return{x:function(){return t.mulStrict(e.exp())}}})},e.log=function(e){o.assertArgumentsAreTensors({x:e},"log");return a.ENV.engine.runKernel(function(t){return t.log(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.toFloat())}}})},e.log1p=function(e){o.assertArgumentsAreTensors({x:e},"log1p");return a.ENV.engine.runKernel(function(t){return t.log1p(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.add(u.scalar(1)))}}})},e.sqrt=function(e){o.assertArgumentsAreTensors({x:e},"sqrt");return a.ENV.engine.runKernel(function(t){return t.sqrt(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.toFloat().sqrt().mul(u.scalar(2)))}}})},e.rsqrt=function(e){o.assertArgumentsAreTensors({x:e},"rsqrt");return a.ENV.engine.runKernel(function(t){return t.rsqrt(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.pow(u.scalar(1.5)).mul(u.scalar(2))).neg()}}})},e.square=function(e){o.assertArgumentsAreTensors({x:e},"square");return a.ENV.engine.runKernel(function(t){return t.square(e)},{x:e},function(t){return{x:function(){return t.mulStrict(e.toFloat().mul(u.scalar(2)))}}})},e.reciprocal=function(e){o.assertArgumentsAreTensors({x:e},"reciprocal");return a.ENV.engine.runKernel(function(t){return t.reciprocal(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.square().neg())}}})},e.abs=function(e){o.assertArgumentsAreTensors({x:e},"abs");return a.ENV.engine.runKernel(function(t){return t.abs(e)},{x:e},function(t){return{x:function(){return t.mulStrict(e.toFloat().step(-1))}}})},e.clipByValue=function(e,t,n){o.assertArgumentsAreTensors({x:e},"clipByValue"),o.assert(t<=n,"Error in clip: min ("+t+") must be less than or equal to max ("+n+").");return a.ENV.engine.runKernel(function(r){return r.clip(e,t,n)},{x:e},function(r){return{x:function(){return r.where(e.greater(u.scalar(t)).logicalAnd(e.less(u.scalar(n))),l.zerosLike(r))}}})},e.relu=function(e){if(o.assertArgumentsAreTensors({x:e},"relu"),"bool"===e.dtype)return e.toInt();return a.ENV.engine.runKernel(function(t){return t.relu(e)},{x:e},function(t){var n=e.step();return{x:function(){return t.mulStrict(n.toFloat())}}})},e.elu=function(e){o.assertArgumentsAreTensors({x:e},"elu");return a.ENV.engine.runKernel(function(t,n){return n(t.elu(e))},{x:e},function(e,t){var n=t[0];return{x:function(){return a.ENV.engine.runKernel(function(t){return t.eluDer(e,n)},{dy:e,y:n})}}})},e.selu=function(e){o.assertArgumentsAreTensors({x:e},"selu");return a.ENV.engine.runKernel(function(t){return t.selu(e)},{x:e},function(t){return{x:function(){var n=e.greater(u.scalar(0)),r=u.scalar(c.SELU_SCALEALPHA),i=u.scalar(c.SELU_SCALE),a=t.mul(i),o=t.mul(r).mul(e.toFloat().exp());return u.where(n,a,o)}}})},e.leakyRelu=function(e,t){return void 0===t&&(t=.2),o.assertArgumentsAreTensors({x:e},"leakyRelu"),u.maximum(u.scalar(t).mul(e),e)},e.prelu=function(e,t){o.assertArgumentsAreTensors({x:e,alpha:t},"prelu");var n=u.scalar(0);return u.maximum(n,e).add(t.mul(u.minimum(n,e)))},e.sigmoid=function(e){o.assertArgumentsAreTensors({x:e},"sigmoid");return a.ENV.engine.runKernel(function(t,n){return n(t.sigmoid(e))},{x:e},function(e,t){var n=t[0];return{x:function(){return e.mulStrict(n.mul(u.scalar(1).sub(n)))}}})},e.logSigmoid=function(e){o.assertArgumentsAreTensors({x:e},"logSigmoid");return a.ENV.engine.runKernel(function(t){return t.softplus(e.neg()).neg()},{x:e},function(t){return{x:function(){return t.mulStrict(e.neg().sigmoid())}}})},e.softplus=function(e){o.assertArgumentsAreTensors({x:e},"softplus");return a.ENV.engine.runKernel(function(t){return t.softplus(e)},{x:e},function(t){return{x:function(){return t.mulStrict(e.sigmoid())}}})},e.sin=function(e){o.assertArgumentsAreTensors({x:e},"sin");return a.ENV.engine.runKernel(function(t){return t.sin(e)},{x:e},function(t){return{x:function(){return e.toFloat().cos().mulStrict(t)}}})},e.cos=function(e){o.assertArgumentsAreTensors({x:e},"cos");return a.ENV.engine.runKernel(function(t){return t.cos(e)},{x:e},function(t){return{x:function(){return e.toFloat().sin().neg().mulStrict(t)}}})},e.tan=function(e){o.assertArgumentsAreTensors({x:e},"tan");return a.ENV.engine.runKernel(function(t){return t.tan(e)},{x:e},function(t){return{x:function(){return t.divStrict(e.cos().square())}}})},e.asin=function(t){o.assertArgumentsAreTensors({x:t},"asin");return a.ENV.engine.runKernel(function(e){return e.asin(t)},{x:t},function(n){return{x:function(){return n.divStrict(e.sqrt(u.scalar(1).sub(t.toFloat().square())))}}})},e.acos=function(t){o.assertArgumentsAreTensors({x:t},"acos");return a.ENV.engine.runKernel(function(e){return e.acos(t)},{x:t},function(n){return{x:function(){return n.divStrict(e.sqrt(u.scalar(1).sub(t.toFloat().square()))).neg()}}})},e.atan=function(e){o.assertArgumentsAreTensors({x:e},"atan");return a.ENV.engine.runKernel(function(t){return t.atan(e)},{x:e},function(t){return{x:function(){return t.divStrict(u.scalar(1).add(e.toFloat().square()))}}})},e.sinh=function(e){o.assertArgumentsAreTensors({x:e},"sinh");return a.ENV.engine.runKernel(function(t){return t.sinh(e)},{x:e},function(t){return{x:function(){return e.toFloat().cosh().mulStrict(t)}}})},e.cosh=function(e){o.assertArgumentsAreTensors({x:e},"cosh");return a.ENV.engine.runKernel(function(t){return t.cosh(e)},{x:e},function(t){return{x:function(){return e.toFloat().sinh().mulStrict(t)}}})},e.tanh=function(e){o.assertArgumentsAreTensors({x:e},"tanh");return a.ENV.engine.runKernel(function(t,n){return n(t.tanh(e))},{x:e},function(e,t){var n=t[0];return{x:function(){return u.scalar(1).sub(n.square()).mulStrict(e)}}})},e.asinh=function(t){o.assertArgumentsAreTensors({x:t},"asinh");return a.ENV.engine.runKernel(function(e){return e.asinh(t)},{x:t},function(n){return{x:function(){return n.divStrict(e.sqrt(u.scalar(1).add(t.toFloat().square())))}}})},e.acosh=function(t){o.assertArgumentsAreTensors({x:t},"acosh");return a.ENV.engine.runKernel(function(e){return e.acosh(t)},{x:t},function(n){return{x:function(){return n.divStrict(e.sqrt(t.toFloat().square().sub(u.scalar(1))))}}})},e.atanh=function(e){o.assertArgumentsAreTensors({x:e},"atanh");return a.ENV.engine.runKernel(function(t){return t.atanh(e)},{x:e},function(t){return{x:function(){return t.divStrict(u.scalar(1).sub(e.toFloat().square()))}}})},e.erf=function(e){o.assert("int32"===e.dtype||"float32"===e.dtype,"Input dtype must be `int32` or `float32`."),"int32"===e.dtype&&(e=e.toFloat());return a.ENV.engine.runKernel(function(t){return t.erf(e)},{x:e},function(t){return{x:function(){return t.mulStrict(u.scalar(2/Math.sqrt(Math.PI)).mul(e.square().neg().exp()))}}})},e.step=function(e,t){void 0===t&&(t=0),o.assertArgumentsAreTensors({x:e},"step");return a.ENV.engine.runKernel(function(n){return n.step(e,t)},{x:e},function(e){return{x:function(){return u.zerosLike(e)}}})},r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"neg",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"ceil",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"floor",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"sign",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"round",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"exp",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"expm1",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"log",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"log1p",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"sqrt",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"rsqrt",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"square",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"reciprocal",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"abs",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"clipByValue",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"relu",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"elu",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"selu",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"leakyRelu",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"prelu",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"sigmoid",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"logSigmoid",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"softplus",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"sin",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"cos",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"tan",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"asin",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"acos",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"atan",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"sinh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"cosh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"tanh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"asinh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"acosh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"atanh",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"erf",null),r([i.doc({heading:"Operations",subheading:"Basic math"}),s.operation],e,"step",null),e}();n.UnaryOps=p},{"../doc":3,"../environment":5,"../util":95,"./operation":65,"./ops":66,"./selu_util":72}],78:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n,r){void 0===r&&(r=1e-8);var i=e.call(this)||this;return i.accumulatedGrads={},i.accumulatedUpdates={},i.c=a.keep(o.scalar(-t)),i.epsilon=a.keep(o.scalar(r)),i.rho=a.keep(o.scalar(n)),i.oneMinusRho=a.keep(o.scalar(1-n)),i}return r(t,e),t.prototype.applyGradients=function(e){var t=this,n=this;for(var r in e)!function(r){var s=i.ENV.engine.registeredVariables[r];null==n.accumulatedGrads[r]&&a.tidy(function(){t.accumulatedGrads[r]=o.zerosLike(s).variable(!1)});null==n.accumulatedUpdates[r]&&a.tidy(function(){t.accumulatedUpdates[r]=o.zerosLike(s).variable(!1)});var u=e[r],l=n.accumulatedGrads[r],c=n.accumulatedUpdates[r];a.tidy(function(){var e=t.rho.mul(l).add(t.oneMinusRho.mul(u.square())),n=c.add(t.epsilon).sqrt().div(l.add(t.epsilon).sqrt()).mul(u),i=t.rho.mul(c).add(t.oneMinusRho.mul(n.square()));t.accumulatedGrads[r].assign(e),t.accumulatedUpdates[r].assign(i);var a=t.c.mul(n).add(s);s.assign(a)})}(r)},t.prototype.dispose=function(){var e=this;this.c.dispose(),this.epsilon.dispose(),this.rho.dispose(),this.oneMinusRho.dispose(),null!=this.accumulatedUpdates&&(Object.keys(this.accumulatedUpdates).forEach(function(t){return e.accumulatedUpdates[t].dispose()}),Object.keys(this.accumulatedGrads).forEach(function(t){return e.accumulatedGrads[t].dispose()}))},t}(e("./optimizer").Optimizer);n.AdadeltaOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],79:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n){void 0===n&&(n=.1);var r=e.call(this)||this;return r.learningRate=t,r.initialAccumulatorValue=n,r.accumulatedGrads={},r.c=a.keep(o.scalar(-t)),r.epsilon=a.keep(o.scalar(1e-8)),r}return r(t,e),t.prototype.applyGradients=function(e){var t=this,n=this;for(var r in e)!function(r){var s=i.ENV.engine.registeredVariables[r];null==n.accumulatedGrads[r]&&a.tidy(function(){t.accumulatedGrads[r]=o.fill(s.shape,t.initialAccumulatorValue).variable(!1)});var u=e[r],l=n.accumulatedGrads[r];a.tidy(function(){var e=l.add(u.square());t.accumulatedGrads[r].assign(e);var n=t.c.mul(u.div(e.add(t.epsilon).sqrt())).add(s);s.assign(n)})}(r)},t.prototype.dispose=function(){var e=this;this.epsilon.dispose(),this.c.dispose(),null!=this.accumulatedGrads&&Object.keys(this.accumulatedGrads).forEach(function(t){return e.accumulatedGrads[t].dispose()})},t}(e("./optimizer").Optimizer);n.AdagradOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],80:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n,r,i){void 0===i&&(i=1e-8);var s=e.call(this)||this;return s.learningRate=t,s.accumulatedFirstMoment={},s.accumulatedSecondMoment={},s.c=a.keep(o.scalar(-t)),s.eps=a.keep(o.scalar(i)),s.beta1=a.keep(o.scalar(n)),s.beta2=a.keep(o.scalar(r)),a.tidy(function(){s.accBeta1=o.scalar(n).variable(),s.accBeta2=o.scalar(r).variable()}),s.oneMinusBeta1=a.keep(o.scalar(1-n)),s.oneMinusBeta2=a.keep(o.scalar(1-r)),s.one=a.keep(o.scalar(1)),s}return r(t,e),t.prototype.applyGradients=function(e){var t=this;a.tidy(function(){var n=t.one.sub(t.accBeta1),r=t.one.sub(t.accBeta2);for(var a in e){var s=i.ENV.engine.registeredVariables[a];if(null==t.accumulatedFirstMoment[a]){u=!1;t.accumulatedFirstMoment[a]=o.zerosLike(s).variable(u)}if(null==t.accumulatedSecondMoment[a]){var u=!1;t.accumulatedSecondMoment[a]=o.zerosLike(s).variable(u)}var l=e[a],c=t.accumulatedFirstMoment[a],p=t.accumulatedSecondMoment[a],h=t.beta1.mul(c).add(t.oneMinusBeta1.mul(l)),d=t.beta2.mul(p).add(t.oneMinusBeta2.mul(l.square())),f=h.div(n),g=d.div(r);t.accumulatedFirstMoment[a].assign(h),t.accumulatedSecondMoment[a].assign(d);var m=t.c.mul(f.div(t.eps.add(g.sqrt()))).add(s);s.assign(m)}t.accBeta1.assign(t.accBeta1.mul(t.beta1)),t.accBeta2.assign(t.accBeta2.mul(t.beta2))})},t.prototype.dispose=function(){var e=this;this.c.dispose(),this.eps.dispose(),this.beta1.dispose(),this.beta2.dispose(),this.accBeta1.dispose(),this.accBeta2.dispose(),this.oneMinusBeta1.dispose(),this.oneMinusBeta2.dispose(),this.one.dispose(),null!=this.accumulatedFirstMoment&&Object.keys(this.accumulatedFirstMoment).forEach(function(t){return e.accumulatedFirstMoment[t].dispose()}),null!=this.accumulatedSecondMoment&&Object.keys(this.accumulatedSecondMoment).forEach(function(t){return e.accumulatedSecondMoment[t].dispose()})},t}(e("./optimizer").Optimizer);n.AdamOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],81:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n,r,i,s){void 0===i&&(i=1e-8),void 0===s&&(s=0);var u=e.call(this)||this;return u.learningRate=t,u.accumulatedFirstMoment={},u.accumulatedWeightedInfNorm={},u.c=a.keep(o.scalar(-t)),u.eps=a.keep(o.scalar(i)),u.beta1=a.keep(o.scalar(n)),u.beta2=a.keep(o.scalar(r)),u.decay=a.keep(o.scalar(s)),a.tidy(function(){u.iteration=o.scalar(0).variable(),u.accBeta1=o.scalar(n).variable()}),u.oneMinusBeta1=a.keep(o.scalar(1-n)),u.one=a.keep(o.scalar(1)),u}return r(t,e),t.prototype.applyGradients=function(e){var t=this;a.tidy(function(){var n=t.one.sub(t.accBeta1),r=t.c.div(t.one.add(t.decay.mul(t.iteration)));for(var a in e){var s=i.ENV.engine.registeredVariables[a];if(null==t.accumulatedFirstMoment[a]){u=!1;t.accumulatedFirstMoment[a]=o.zerosLike(s).variable(u)}if(null==t.accumulatedWeightedInfNorm[a]){var u=!1;t.accumulatedWeightedInfNorm[a]=o.zerosLike(s).variable(u)}var l=e[a],c=t.accumulatedFirstMoment[a],p=t.accumulatedWeightedInfNorm[a],h=t.beta1.mul(c).add(t.oneMinusBeta1.mul(l)),d=t.beta2.mul(p),f=l.abs(),g=d.maximum(f);t.accumulatedFirstMoment[a].assign(h),t.accumulatedWeightedInfNorm[a].assign(g);var m=r.div(n).mul(h.div(t.eps.add(g))).add(s);s.assign(m)}t.iteration.assign(t.iteration.add(t.one)),t.accBeta1.assign(t.accBeta1.mul(t.beta1))})},t.prototype.dispose=function(){var e=this;this.c.dispose(),this.eps.dispose(),this.accBeta1.dispose(),this.beta1.dispose(),this.beta2.dispose(),this.oneMinusBeta1.dispose(),this.decay.dispose(),this.iteration.dispose(),this.one.dispose(),null!=this.accumulatedFirstMoment&&Object.keys(this.accumulatedFirstMoment).forEach(function(t){return e.accumulatedFirstMoment[t].dispose()}),null!=this.accumulatedWeightedInfNorm&&Object.keys(this.accumulatedWeightedInfNorm).forEach(function(t){return e.accumulatedWeightedInfNorm[t].dispose()})},t}(e("./optimizer").Optimizer);n.AdamaxOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],82:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n,r){void 0===r&&(r=!1);var i=e.call(this,t)||this;return i.learningRate=t,i.momentum=n,i.useNesterov=r,i.m=o.scalar(i.momentum),i.accumulations={},i}return r(t,e),t.prototype.applyGradients=function(e){var t=this,n=this;for(var r in e)!function(r){var s=i.ENV.engine.registeredVariables[r];null==n.accumulations[r]&&a.tidy(function(){t.accumulations[r]=o.zerosLike(s).variable(!1)});var u=n.accumulations[r],l=e[r];a.tidy(function(){var e,n=t.m.mul(u).add(l);e=t.useNesterov?t.c.mul(l.add(n.mul(t.m))).add(s):t.c.mul(n).add(s),t.accumulations[r].assign(n),s.assign(e)})}(r)},t.prototype.dispose=function(){if(e.prototype.dispose.call(this),this.m.dispose(),null!=this.accumulations)for(var t in this.accumulations)this.accumulations[t].dispose()},t.prototype.setMomentum=function(e){this.momentum=e},t}(e("./sgd_optimizer").SGDOptimizer);n.MomentumOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./sgd_optimizer":86}],83:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("../globals"),o=function(){function e(){}return e.prototype.minimize=function(e,t,n){void 0===t&&(t=!1);var r=this.computeGradients(e,n),i=r.value,a=r.grads;return this.applyGradients(a),Object.keys(a).forEach(function(e){return a[e].dispose()}),t?i:(i.dispose(),null)},e.prototype.computeGradients=function(e,t){return a.variableGrads(e,t)},r([i.doc({heading:"Training",subheading:"Optimizers"})],e.prototype,"minimize",null),e=r([i.doc({heading:"Training",subheading:"Classes",namespace:"train"})],e)}();n.Optimizer=o},{"../doc":3,"../globals":6}],84:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("../doc"),a=e("./adadelta_optimizer"),o=e("./adagrad_optimizer"),s=e("./adam_optimizer"),u=e("./adamax_optimizer"),l=e("./momentum_optimizer"),c=e("./rmsprop_optimizer"),p=e("./sgd_optimizer"),h=function(){function e(){}return e.sgd=function(e){return new p.SGDOptimizer(e)},e.momentum=function(e,t,n){return void 0===n&&(n=!1),new l.MomentumOptimizer(e,t,n)},e.rmsprop=function(e,t,n,r,i){return void 0===t&&(t=.9),void 0===n&&(n=0),void 0===r&&(r=1e-8),void 0===i&&(i=!1),new c.RMSPropOptimizer(e,t,n,r,i)},e.adam=function(e,t,n,r){return void 0===e&&(e=.001),void 0===t&&(t=.9),void 0===n&&(n=.999),void 0===r&&(r=1e-8),new s.AdamOptimizer(e,t,n,r)},e.adadelta=function(e,t,n){return void 0===e&&(e=.001),void 0===t&&(t=.95),void 0===n&&(n=1e-8),new a.AdadeltaOptimizer(e,t,n)},e.adamax=function(e,t,n,r,i){return void 0===e&&(e=.002),void 0===t&&(t=.9),void 0===n&&(n=.999),void 0===r&&(r=1e-8),void 0===i&&(i=0),new u.AdamaxOptimizer(e,t,n,r,i)},e.adagrad=function(e,t){return void 0===t&&(t=.1),new o.AdagradOptimizer(e,t)},r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"sgd",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"momentum",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"rmsprop",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"adam",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"adadelta",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"adamax",null),r([i.doc({heading:"Training",subheading:"Optimizers",namespace:"train"})],e,"adagrad",null),e}();n.OptimizerConstructors=h},{"../doc":3,"./adadelta_optimizer":78,"./adagrad_optimizer":79,"./adam_optimizer":80,"./adamax_optimizer":81,"./momentum_optimizer":82,"./rmsprop_optimizer":85,"./sgd_optimizer":86}],85:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t,n,r,i,s){void 0===n&&(n=.9),void 0===r&&(r=0),void 0===i&&(i=1e-8),void 0===s&&(s=!1);var u=e.call(this)||this;return u.learningRate=t,u.accumulatedMeanSquares={},u.accumulatedMeanGrads={},u.accumulatedMoments={},u.c=a.keep(o.scalar(t)),u.epsilon=a.keep(o.scalar(i)),u.decay=a.keep(o.scalar(n)),u.momentum=a.keep(o.scalar(r)),u.oneMinusDecay=a.keep(o.scalar(1-n)),u.centered=s,u}return r(t,e),t.prototype.applyGradients=function(e){var t=this,n=this;for(var r in e)!function(r){var s=i.ENV.engine.registeredVariables[r];null==n.accumulatedMeanSquares[r]&&a.tidy(function(){t.accumulatedMeanSquares[r]=o.zerosLike(s).variable(!1)});null==n.accumulatedMeanGrads[r]&&n.centered&&a.tidy(function(){t.accumulatedMeanGrads[r]=o.zerosLike(s).variable(!1)});null==n.accumulatedMoments[r]&&a.tidy(function(){t.accumulatedMoments[r]=o.zerosLike(s).variable(!1)});var u=n.accumulatedMeanSquares[r],l=n.accumulatedMeanGrads[r],c=n.accumulatedMoments[r],p=e[r];a.tidy(function(){var e=t.decay.mul(u).add(t.oneMinusDecay.mul(p.square()));if(t.centered){var n=t.decay.mul(l).add(t.oneMinusDecay.mul(p)),i=t.momentum.mul(c).add(t.c.mul(p).div(e.sub(n.square().add(t.epsilon)).sqrt()));t.accumulatedMeanSquares[r].assign(e),t.accumulatedMeanGrads[r].assign(n),t.accumulatedMoments[r].assign(i),o=s.sub(i),s.assign(o)}else{var a=t.decay.mul(u).add(t.oneMinusDecay.mul(p.square())),i=t.momentum.mul(c).add(t.c.mul(p).div(a.add(t.epsilon).sqrt()));t.accumulatedMeanSquares[r].assign(a),t.accumulatedMoments[r].assign(i);var o=s.sub(i);s.assign(o)}})}(r)},t.prototype.dispose=function(){var e=this;this.c.dispose(),this.epsilon.dispose(),this.decay.dispose(),this.momentum.dispose(),this.oneMinusDecay.dispose(),null!=this.accumulatedMeanSquares&&Object.keys(this.accumulatedMeanSquares).forEach(function(t){return e.accumulatedMeanSquares[t].dispose()}),null!=this.accumulatedMeanGrads&&this.centered&&Object.keys(this.accumulatedMeanGrads).forEach(function(t){return e.accumulatedMeanGrads[t].dispose()}),null!=this.accumulatedMoments&&Object.keys(this.accumulatedMoments).forEach(function(t){return e.accumulatedMoments[t].dispose()})},t}(e("./optimizer").Optimizer);n.RMSPropOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],86:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../environment"),a=e("../globals"),o=e("../ops/ops"),s=function(e){function t(t){var n=e.call(this)||this;return n.learningRate=t,n.setLearningRate(t),n}return r(t,e),t.prototype.applyGradients=function(e){var t=this;Object.keys(e).forEach(function(n){var r=e[n],o=i.ENV.engine.registeredVariables[n];a.tidy(function(){var e=t.c.mul(r).add(o);o.assign(e)})})},t.prototype.setLearningRate=function(e){this.learningRate=e,null!=this.c&&this.c.dispose(),this.c=a.keep(o.scalar(-e))},t.prototype.dispose=function(){this.c.dispose()},t}(e("./optimizer").Optimizer);n.SGDOptimizer=s},{"../environment":5,"../globals":6,"../ops/ops":66,"./optimizer":83}],87:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./util"),i=function(){function e(e,t){this.backendTimer=e,this.logger=t,null==t&&(this.logger=new a)}return e.prototype.profileKernel=function(e,t){var n,i=this,a=this.backendTimer.time(function(){n=t()}),o=n.dataSync();return r.checkForNaN(o,n.dtype,e),a.then(function(t){i.logger.logKernelProfile(e,n,o,t.kernelMs)}),n},e}();n.Profiler=i;var a=function(){function e(){}return e.prototype.logKernelProfile=function(e,t,n,i){var a=r.rightPad(i+"ms",9),o=r.rightPad(e,25),s=t.rank,u=t.size,l=r.rightPad(t.shape.toString(),14);console.log("%c"+o+"\t%c"+a+"\t%c"+s+"D "+l+"\t%c"+u,"font-weight:bold","color:red","color:blue","color: orange")},e}();n.Logger=a},{"./util":95}],88:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./util");n.getFilteredNodesXToY=function(e,t,n){for(var r={},i={},a=0;a<t.length;a++)r[t[a].id]=!0;for(a=0;a<e.length;a++){h=(g=e[a]).inputs;for(var o in h){for(var s=h[o],u=!1,l=0;l<t.length;l++)if(r[s.id]){r[g.output.id]=!0,u=!0,i[g.id]=!0;break}if(u)break}}var c={};c[n.id]=!0;for(var p={},a=e.length-1;a>=0;a--){var h=(g=e[a]).inputs,d=[];for(d.push(g.output),l=0;l<d.length;l++)if(c[d[l].id]){for(var o in h)c[h[o].id]=!0,p[g.id]=!0;break}}for(var f=[],a=0;a<e.length;a++){var g=e[a];if(i[g.id]&&p[g.id]){var m={};for(var o in g.inputs){var v=g.inputs[o];r[v.id]&&(m[o]=v)}var y=Object.assign({},g);y.inputs=m,y.output=g.output,f.push(y)}}return f},n.backpropagateGradients=function(e,t){for(var n=t.length-1;n>=0;n--){var i=t[n],a=e[i.output.id];if(null==i.gradient)throw new Error("Cannot compute gradient: gradient function not found for "+i.name+".");var o=i.gradient(a);for(var s in i.inputs){if(!(s in o))throw new Error("Cannot backprop through input "+s+". Available gradients found: "+Object.keys(o)+".");var u=o[s](),l=i.inputs[s];if(!r.arraysEqual(u.shape,l.shape))throw new Error("Error in gradient for op "+i.name+". The gradient of input '"+s+"' has shape '"+u.shape+"', which does not match the shape of the input '"+l.shape+"'");if(null==e[l.id])e[l.id]=u;else{var c=e[l.id];e[l.id]=c.add(u),c.dispose()}}}}},{"./util":95}],89:[function(e,t,n){"use strict";function r(e){var t=e.length;if(t<2)return[];var n=new Array(t-1);n[t-2]=e[t-1];for(var r=t-3;r>=0;--r)n[r]=n[r+1]*e[r+1];return n}var i=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),a=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o},o=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},s=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var u=e("./doc"),l=e("./environment"),c=e("./ops/ops"),p=e("./tensor_util"),h=e("./util"),d=function(){function e(e,t,n){if(this.dtype=t,null!=n){var i=n.length,a=h.sizeFromShape(e);h.assert(i===a,"Length of values '"+i+"' does not match the size inferred by the shape '"+a+"'")}this.shape=e.slice(),this.values=n||h.getTypedArrayFromDType(t,h.sizeFromShape(e)),this.strides=r(e),this.size=h.sizeFromShape(e)}return e.prototype.set=function(e){for(var t=[],n=1;n<arguments.length;n++)t[n-1]=arguments[n];0===t.length&&(t=[0]),h.assert(t.length===this.rank,"The number of provided coordinates ("+t.length+") must match the rank ("+this.rank+")");var r=this.locToIndex(t);this.values[r]=e},e.prototype.get=function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];0===e.length&&(e=[0]);for(var n=e[e.length-1],r=0;r<e.length-1;++r)n+=this.strides[r]*e[r];return this.values[n]},e.prototype.locToIndex=function(e){if(0===this.rank)return 0;if(1===this.rank)return e[0];for(var t=e[e.length-1],n=0;n<e.length-1;++n)t+=this.strides[n]*e[n];return t},e.prototype.indexToLoc=function(e){if(0===this.rank)return[];if(1===this.rank)return[e];for(var t=new Array(this.shape.length),n=0;n<t.length-1;++n)t[n]=Math.floor(e/this.strides[n]),e-=t[n]*this.strides[n];return t[t.length-1]=e,t},Object.defineProperty(e.prototype,"rank",{get:function(){return this.shape.length},enumerable:!0,configurable:!0}),e.prototype.toTensor=function(){return f.make(this.shape,{values:this.values},this.dtype)},a([u.doc({heading:"Tensors",subheading:"Creation"})],e.prototype,"set",null),a([u.doc({heading:"Tensors",subheading:"Creation"})],e.prototype,"get",null),a([u.doc({heading:"Tensors",subheading:"Creation"})],e.prototype,"toTensor",null),e=a([u.doc({heading:"Tensors",subheading:"Classes"})],e)}();n.TensorBuffer=d;var f=function(){function e(e,n,i,a){this.isDisposed=!1,this.size=h.sizeFromShape(e),null!=i&&h.assert(this.size===i.length,"Constructing tensor of shape ("+this.size+") should match the length of values ("+i.length+")"),this.shape=e.slice(),this.dtype=n||"float32",this.strides=r(e),this.dataId=null!=a?a:{},this.id=t.nextId++,this.rankType=this.rank<5?this.rank.toString():"higher",l.ENV.engine.registerTensor(this),null!=i&&l.ENV.engine.write(this.dataId,i)}return t=e,e.make=function(e,n,r){return new t(e,r,n.values,n.dataId)},e.prototype.flatten=function(){return this.throwIfDisposed(),this.as1D()},e.prototype.asScalar=function(){return this.throwIfDisposed(),h.assert(1===this.size,"The array must have only 1 element."),this.reshape([])},e.prototype.as1D=function(){return this.throwIfDisposed(),this.reshape([this.size])},e.prototype.as2D=function(e,t){return this.throwIfDisposed(),this.reshape([e,t])},e.prototype.as3D=function(e,t,n){return this.throwIfDisposed(),this.reshape([e,t,n])},e.prototype.as4D=function(e,t,n,r){return this.throwIfDisposed(),this.reshape([e,t,n,r])},e.prototype.asType=function(e){return this.throwIfDisposed(),c.cast(this,e)},Object.defineProperty(e.prototype,"rank",{get:function(){return this.shape.length},enumerable:!0,configurable:!0}),e.prototype.get=function(){for(var e=[],t=0;t<arguments.length;t++)e[t]=arguments[t];h.assert(e.length===this.rank,"Number of coordinates in get() must match the rank of the tensor"),this.throwIfDisposed(),0===e.length&&(e=[0]);for(var n=e[e.length-1],r=0;r<e.length-1;++r)n+=this.strides[r]*e[r];return this.dataSync()[n]},e.prototype.buffer=function(){return c.buffer(this.shape,this.dtype,this.dataSync())},e.prototype.data=function(){return o(this,void 0,void 0,function(){return s(this,function(e){return this.throwIfDisposed(),[2,l.ENV.engine.read(this.dataId)]})})},e.prototype.dataSync=function(){return this.throwIfDisposed(),l.ENV.engine.readSync(this.dataId)},e.prototype.dispose=function(){this.isDisposed||(this.isDisposed=!0,l.ENV.engine.disposeTensor(this))},e.prototype.throwIfDisposed=function(){if(this.isDisposed)throw new Error("Tensor is disposed.")},e.prototype.toFloat=function(){return this.asType("float32")},e.prototype.toInt=function(){return this.asType("int32")},e.prototype.toBool=function(){return this.asType("bool")},e.prototype.print=function(e){return void 0===e&&(e=!1),c.print(this,e)},e.prototype.reshape=function(e){return this.throwIfDisposed(),c.reshape(this,e)},e.prototype.reshapeAs=function(e){return this.throwIfDisposed(),this.reshape(e.shape)},e.prototype.expandDims=function(e){return void 0===e&&(e=0),c.expandDims(this,e)},e.prototype.squeeze=function(e){return this.throwIfDisposed(),c.squeeze(this,e)},e.prototype.clone=function(){return this.throwIfDisposed(),c.clone(this)},e.prototype.toString=function(e){return void 0===e&&(e=!1),p.tensorToString(this,e)},e.prototype.tile=function(e){return this.throwIfDisposed(),c.tile(this,e)},e.prototype.gather=function(e,t){return void 0===t&&(t=0),this.throwIfDisposed(),c.gather(this,e,t)},e.prototype.matMul=function(e,t,n){return void 0===t&&(t=!1),void 0===n&&(n=!1),this.throwIfDisposed(),c.matMul(this,e,t,n)},e.prototype.norm=function(e,t,n){return void 0===e&&(e="euclidean"),void 0===t&&(t=null),void 0===n&&(n=!1),this.throwIfDisposed(),c.norm(this,e,t,n)},e.prototype.slice=function(e,t){return this.throwIfDisposed(),c.slice(this,e,t)},e.prototype.reverse=function(e){return this.throwIfDisposed(),c.reverse(this,e)},e.prototype.concat=function(e,t){return void 0===t&&(t=0),this.throwIfDisposed(),c.concat([this,e],t)},e.prototype.stack=function(e,t){return void 0===t&&(t=0),c.stack([this,e],t)},e.prototype.pad=function(e,t){return void 0===t&&(t=0),c.pad(this,e,t)},e.prototype.batchNormalization=function(e,t,n,r,i){return void 0===n&&(n=.001),this.throwIfDisposed(),c.batchNormalization(this,e,t,n,r,i)},e.prototype.logSumExp=function(e,t){return void 0===e&&(e=null),void 0===t&&(t=!1),this.throwIfDisposed(),c.logSumExp(this,e,t)},e.prototype.sum=function(e,t){return void 0===e&&(e=null),void 0===t&&(t=!1),this.throwIfDisposed(),c.sum(this,e,t)},e.prototype.mean=function(e,t){return void 0===e&&(e=null),void 0===t&&(t=!1),this.throwIfDisposed(),c.mean(this,e,t)},e.prototype.min=function(e,t){return void 0===e&&(e=null),void 0===t&&(t=!1),this.throwIfDisposed(),c.min(this,e,t)},e.prototype.max=function(e,t){return void 0===e&&(e=null),void 0===t&&(t=!1),this.throwIfDisposed(),c.max(this,e,t)},e.prototype.argMin=function(e){return void 0===e&&(e=null),this.throwIfDisposed(),c.argMin(this,e)},e.prototype.argMax=function(e){return void 0===e&&(e=null),this.throwIfDisposed(),c.argMax(this,e)},e.prototype.cast=function(e){return this.throwIfDisposed(),c.cast(this,e)},e.prototype.add=function(e){return this.throwIfDisposed(),c.add(this,e)},e.prototype.addStrict=function(e){return this.throwIfDisposed(),c.addStrict(this,e)},e.prototype.sub=function(e){return this.throwIfDisposed(),c.sub(this,e)},e.prototype.subStrict=function(e){return this.throwIfDisposed(),c.subStrict(this,e)},e.prototype.pow=function(e){return this.throwIfDisposed(),c.pow(this,e)},e.prototype.powStrict=function(e){return this.throwIfDisposed(),c.powStrict(this,e)},e.prototype.mul=function(e){return this.throwIfDisposed(),c.mul(this,e)},e.prototype.mulStrict=function(e){return this.throwIfDisposed(),c.mulStrict(this,e)},e.prototype.div=function(e){return this.throwIfDisposed(),c.div(this,e)},e.prototype.divStrict=function(e){return this.throwIfDisposed(),c.divStrict(this,e)},e.prototype.minimum=function(e){return this.throwIfDisposed(),c.minimum(this,e)},e.prototype.minimumStrict=function(e){return this.throwIfDisposed(),c.minimumStrict(this,e)},e.prototype.maximum=function(e){return this.throwIfDisposed(),c.maximum(this,e)},e.prototype.maximumStrict=function(e){return this.throwIfDisposed(),c.maximumStrict(this,e)},e.prototype.mod=function(e){return this.throwIfDisposed(),c.mod(this,e)},e.prototype.modStrict=function(e){return this.throwIfDisposed(),c.modStrict(this,e)},e.prototype.squaredDifference=function(e){return this.throwIfDisposed(),c.squaredDifference(this,e)},e.prototype.squaredDifferenceStrict=function(e){return this.throwIfDisposed(),c.squaredDifferenceStrict(this,e)},e.prototype.transpose=function(e){return this.throwIfDisposed(),c.transpose(this,e)},e.prototype.notEqual=function(e){return this.throwIfDisposed(),c.notEqual(this,e)},e.prototype.notEqualStrict=function(e){return this.throwIfDisposed(),c.notEqualStrict(this,e)},e.prototype.less=function(e){return this.throwIfDisposed(),c.less(this,e)},e.prototype.lessStrict=function(e){return this.throwIfDisposed(),c.lessStrict(this,e)},e.prototype.equal=function(e){return this.throwIfDisposed(),c.equal(this,e)},e.prototype.equalStrict=function(e){return this.throwIfDisposed(),c.equalStrict(this,e)},e.prototype.lessEqual=function(e){return this.throwIfDisposed(),c.lessEqual(this,e)},e.prototype.lessEqualStrict=function(e){return this.throwIfDisposed(),c.lessEqualStrict(this,e)},e.prototype.greater=function(e){return this.throwIfDisposed(),c.greater(this,e)},e.prototype.greaterStrict=function(e){return this.throwIfDisposed(),c.greaterStrict(this,e)},e.prototype.greaterEqual=function(e){return this.throwIfDisposed(),c.greaterEqual(this,e)},e.prototype.greaterEqualStrict=function(e){return this.throwIfDisposed(),c.greaterEqualStrict(this,e)},e.prototype.logicalAnd=function(e){return this.throwIfDisposed(),c.logicalAnd(this,e)},e.prototype.logicalOr=function(e){return this.throwIfDisposed(),c.logicalOr(this,e)},e.prototype.logicalNot=function(){return this.throwIfDisposed(),c.logicalNot(this)},e.prototype.logicalXor=function(e){return this.throwIfDisposed(),c.logicalXor(this,e)},e.prototype.where=function(e,t){return this.throwIfDisposed(),c.where(e,this,t)},e.prototype.neg=function(){return this.throwIfDisposed(),c.neg(this)},e.prototype.ceil=function(){return this.throwIfDisposed(),c.ceil(this)},e.prototype.floor=function(){return this.throwIfDisposed(),c.floor(this)},e.prototype.sign=function(){return this.throwIfDisposed(),c.sign(this)},e.prototype.exp=function(){return this.throwIfDisposed(),c.exp(this)},e.prototype.expm1=function(){return this.throwIfDisposed(),c.expm1(this)},e.prototype.log=function(){return this.throwIfDisposed(),c.log(this)},e.prototype.log1p=function(){return this.throwIfDisposed(),c.log1p(this)},e.prototype.sqrt=function(){return this.throwIfDisposed(),c.sqrt(this)},e.prototype.rsqrt=function(){return this.throwIfDisposed(),c.rsqrt(this)},e.prototype.square=function(){return this.throwIfDisposed(),c.square(this)},e.prototype.reciprocal=function(){return this.throwIfDisposed(),c.reciprocal(this)},e.prototype.abs=function(){return this.throwIfDisposed(),c.abs(this)},e.prototype.clipByValue=function(e,t){return this.throwIfDisposed(),c.clipByValue(this,e,t)},e.prototype.relu=function(){return this.throwIfDisposed(),c.relu(this)},e.prototype.elu=function(){return this.throwIfDisposed(),c.elu(this)},e.prototype.selu=function(){return this.throwIfDisposed(),c.selu(this)},e.prototype.leakyRelu=function(e){return void 0===e&&(e=.2),this.throwIfDisposed(),c.leakyRelu(this,e)},e.prototype.prelu=function(e){return this.throwIfDisposed(),c.prelu(this,e)},e.prototype.sigmoid=function(){return this.throwIfDisposed(),c.sigmoid(this)},e.prototype.logSigmoid=function(){return this.throwIfDisposed(),c.logSigmoid(this)},e.prototype.softplus=function(){return this.throwIfDisposed(),c.softplus(this)},e.prototype.sin=function(){return this.throwIfDisposed(),c.sin(this)},e.prototype.cos=function(){return this.throwIfDisposed(),c.cos(this)},e.prototype.tan=function(){return this.throwIfDisposed(),c.tan(this)},e.prototype.asin=function(){return this.throwIfDisposed(),c.asin(this)},e.prototype.acos=function(){return this.throwIfDisposed(),c.acos(this)},e.prototype.atan=function(){return this.throwIfDisposed(),c.atan(this)},e.prototype.sinh=function(){return this.throwIfDisposed(),c.sinh(this)},e.prototype.cosh=function(){return this.throwIfDisposed(),c.cosh(this)},e.prototype.tanh=function(){return this.throwIfDisposed(),c.tanh(this)},e.prototype.asinh=function(){return this.throwIfDisposed(),c.asinh(this)},e.prototype.acosh=function(){return this.throwIfDisposed(),c.acosh(this)},e.prototype.atanh=function(){return this.throwIfDisposed(),c.atanh(this)},e.prototype.erf=function(){return this.throwIfDisposed(),c.erf(this)},e.prototype.step=function(e){return void 0===e&&(e=0),this.throwIfDisposed(),c.step(this,e)},e.prototype.softmax=function(e){return void 0===e&&(e=-1),this.throwIfDisposed(),c.softmax(this,e)},e.prototype.resizeBilinear=function(e,t){return void 0===t&&(t=!1),this.throwIfDisposed(),c.image.resizeBilinear(this,e,t)},e.prototype.resizeNearestNeighbor=function(e,t){return void 0===t&&(t=!1),this.throwIfDisposed(),c.image.resizeNearestNeighbor(this,e,t)},e.prototype.conv1d=function(e,t,n,r,i,a){return void 0===r&&(r="NWC"),void 0===i&&(i=1),this.throwIfDisposed(),c.conv1d(this,e,t,n,r,i,a)},e.prototype.conv2d=function(e,t,n,r,i,a){return void 0===r&&(r="NHWC"),void 0===i&&(i=[1,1]),this.throwIfDisposed(),c.conv2d(this,e,t,n,r,i,a)},e.prototype.conv2dTranspose=function(e,t,n,r,i){return this.throwIfDisposed(),c.conv2dTranspose(this,e,t,n,r,i)},e.prototype.depthwiseConv2D=function(e,t,n,r,i,a){return void 0===r&&(r="NHWC"),void 0===i&&(i=[1,1]),this.throwIfDisposed(),c.depthwiseConv2d(this,e,t,n,r,i,a)},e.prototype.avgPool=function(e,t,n,r){return this.throwIfDisposed(),c.avgPool(this,e,t,n,r)},e.prototype.maxPool=function(e,t,n,r){return this.throwIfDisposed(),c.maxPool(this,e,t,n,r)},e.prototype.localResponseNormalization=function(e,t,n,r){return void 0===e&&(e=5),void 0===t&&(t=1),void 0===n&&(n=1),void 0===r&&(r=.5),c.localResponseNormalization(this,e,t,n,r)},e.prototype.variable=function(e,t,n){return void 0===e&&(e=!0),this.throwIfDisposed(),g.variable(this,e,t,n)},e.nextId=0,a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"flatten",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"asScalar",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"as1D",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"as2D",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"as3D",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"as4D",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"asType",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"buffer",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"data",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"dataSync",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"dispose",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"toFloat",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"toInt",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"toBool",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"print",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"reshape",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"reshapeAs",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"expandDims",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"squeeze",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"clone",null),a([u.doc({heading:"Tensors",subheading:"Classes"})],e.prototype,"toString",null),e=t=a([u.doc({heading:"Tensors",subheading:"Classes"})],e);var t}();n.Tensor=f;var g=function(e){function t(t,r,i){void 0===r&&(r=!0);var a=e.call(this,t.shape,t.dtype,null,t.dataId)||this;return a.trainable=r,a.name=i,null==a.name&&(a.name=n.nextVarId.toString(),n.nextVarId++),l.ENV.engine.registerVariable(a),a}return i(t,e),n=t,t.variable=function(e,t,r,i){return void 0===t&&(t=!0),null!=i&&i!==e.dtype&&(e=e.asType(i)),new n(e,t,r)},t.prototype.assign=function(e){if(e.dtype!==this.dtype)throw new Error("dtype of the new value ("+e.dtype+") and previous value ("+this.dtype+") must match");if(!h.arraysEqual(e.shape,this.shape))throw new Error("shape of the new value ("+e.shape+") and previous value ("+this.shape+") must match");l.ENV.engine.disposeTensor(this),this.dataId=e.dataId,l.ENV.engine.registerTensor(this)},t.nextVarId=0,a([u.doc({heading:"Tensors",subheading:"Classes"})],t.prototype,"assign",null),a([u.doc({heading:"Tensors",subheading:"Creation"})],t,"variable",null),t=n=a([u.doc({heading:"Tensors",subheading:"Classes"})],t);var n}(f);n.Variable=g;var m=g.variable;n.variable=m},{"./doc":3,"./environment":5,"./ops/ops":66,"./tensor_util":90,"./util":95}],90:[function(e,t,n){"use strict";function r(e){var t=e.dataSync(),n=e.size,r=e.strides[e.strides.length-1],a=new Array(r).fill(0);if(e.rank>1)for(var o=0;o<n/r;o++)for(var s=o*r,u=0;u<r;u++)a[u]=Math.max(a[u],i(t[s+u],0).length);return a}function i(e,t){return o.rightPad(parseFloat(e.toFixed(l)).toString(),t)}function a(e,t,n,r,o){void 0===o&&(o=!0);var l=t[0],c=t.length;if(0===c)return[e[0].toString()];if(1===c){if(l>s){var p=Array.from(e.subarray(0,u)),h=Array.from(e.subarray(l-u,l));return["["+p.map(function(e,t){return i(e,r[t])}).join(", ")+", ..., "+h.map(function(e,t){return i(e,r[l-u+t])}).join(", ")+"]"]}return["["+Array.from(e).map(function(e,t){return i(e,r[t])}).join(", ")+"]"]}var d=t.slice(1),f=n.slice(1),g=n[0],m=[];if(l>s){for(x=0;x<u;x++){y=(v=x*g)+g;m.push.apply(m,a(e.subarray(v,y),d,f,r,!1))}m.push("...");for(x=l-u;x<l;x++){y=(v=x*g)+g;m.push.apply(m,a(e.subarray(v,y),d,f,r,x===l-1))}}else for(x=0;x<l;x++){var v=x*g,y=v+g;m.push.apply(m,a(e.subarray(v,y),d,f,r,x===l-1))}var b=2===c?",":"";m[0]="["+m[0]+b;for(x=1;x<m.length-1;x++)m[x]=" "+m[x]+b;for(var w=",\n",x=2;x<c;x++)w+="\n";return m[m.length-1]=" "+m[m.length-1]+"]"+(o?"":w),m}Object.defineProperty(n,"__esModule",{value:!0});var o=e("./util"),s=20,u=3,l=7;n.tensorToString=function(e,t){var n=e.dataSync(),i=r(e),o=a(n,e.shape,e.strides,i),s=["Tensor"];return t&&(s.push("  dtype: "+e.dtype),s.push("  rank: "+e.rank),s.push("  shape: ["+e.shape+"]"),s.push("  values:")),s.push(o.map(function(e){return"    "+e}).join("\n")),s.join("\n")}},{"./util":95}],91:[function(e,t,n){"use strict";function r(e,t,r){if(void 0===r&&(r=n.TEST_EPSILON),e instanceof a.Tensor||t instanceof a.Tensor){if(e instanceof a.Tensor&&t instanceof a.Tensor){if(e.dtype!==t.dtype)throw new Error("Arrays are of different type actual: "+e.dtype+" vs expected: "+t.dtype+".");if(!o.arraysEqual(e.shape,t.shape))throw new Error("Arrays are of different shape actual: "+e.shape+" vs expected: "+t.shape+".")}}else{var s=e.constructor.name,u=t.constructor.name;if(s!==u)throw new Error("Arrays are of different type actual: "+s+" vs expected: "+u)}var l,c;if(l=e instanceof a.Tensor?e.dataSync():e,c=t instanceof a.Tensor?t.dataSync():t,l.length!==c.length)throw new Error("Arrays have different lengths actual: "+l.length+" vs expected: "+c.length+".\nActual:   "+l+".\nExpected: "+c+".");for(var p=0;p<c.length;++p){var h=l[p],d=c[p];if(!i(h,Number(d),r))throw new Error("Arrays differ: actual["+p+"] = "+h+", expected["+p+"] = "+d+".\nActual:   "+l+".\nExpected: "+c+".")}}function i(e,t,n){return!(!isNaN(e)||!isNaN(t))||!(isNaN(e)||isNaN(t)||Math.abs(e-t)>n)}Object.defineProperty(n,"__esModule",{value:!0});var a=e("./tensor"),o=e("./util");n.WEBGL_ENVS={BACKEND:"test-webgl"},n.CPU_ENVS={BACKEND:"test-cpu"},n.ALL_ENVS={},n.TEST_EPSILON=.001,n.expectArraysClose=r,n.expectPromiseToFail=function(e,t){e().then(function(){return t.fail()},function(){return t()})},n.expectArraysEqual=function(e,t){return r(e,t,0)},n.expectNumbersClose=function(e,t,r){if(void 0===r&&(r=n.TEST_EPSILON),!i(e,t,r))throw new Error("Numbers differ: actual === "+e+", expected === "+t)},n.expectValuesInRange=function(e,t,n){var r;r=e instanceof a.Tensor?e.dataSync():e;for(var i=0;i<r.length;i++)if(r[i]<t||r[i]>n)throw new Error("Value out of range:"+r[i]+" low: "+t+", high: "+n)}},{"./tensor":89,"./util":95}],92:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("./doc"),a=e("./environment"),o=e("./util"),s=function(){function e(){}return e.tidy=function(e,t,n){void 0===n&&(n=!1);var r=null;if(null==t){if("function"!=typeof e)throw new Error("Please provide a function to tidy()");t=e}else{if("string"!=typeof e&&!(e instanceof String))throw new Error("When calling with two arguments, the first argument to tidy() must be a string");if("function"!=typeof t)throw new Error("When calling with two arguments, the 2nd argument to tidy() must be a function");r=e}a.ENV.engine.startScope(r,n);var i=t();return i instanceof Promise&&console.error("Cannot return a Promise inside of tidy."),a.ENV.engine.endScope(i,n),i},e.dispose=function(e){o.extractTensorsFromAny(e).forEach(function(e){return e.dispose()})},e.keep=function(e){return a.ENV.engine.keep(e)},e.time=function(e){return a.ENV.engine.time(e)},r([i.doc({heading:"Performance",subheading:"Memory"})],e,"tidy",null),r([i.doc({heading:"Performance",subheading:"Memory"})],e,"keep",null),r([i.doc({heading:"Performance",subheading:"Timing"})],e,"time",null),e}();n.Tracking=s},{"./doc":3,"./environment":5,"./util":95}],93:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./optimizers/adadelta_optimizer"),i=e("./optimizers/adagrad_optimizer"),a=e("./optimizers/adam_optimizer"),o=e("./optimizers/adamax_optimizer"),s=e("./optimizers/momentum_optimizer"),u=e("./optimizers/optimizer_constructors"),l=e("./optimizers/rmsprop_optimizer"),c=e("./optimizers/sgd_optimizer");s.MomentumOptimizer,c.SGDOptimizer,r.AdadeltaOptimizer,i.AdagradOptimizer,l.RMSPropOptimizer,o.AdamaxOptimizer,a.AdamOptimizer,n.train={sgd:u.OptimizerConstructors.sgd,momentum:u.OptimizerConstructors.momentum,adadelta:u.OptimizerConstructors.adadelta,adagrad:u.OptimizerConstructors.adagrad,rmsprop:u.OptimizerConstructors.rmsprop,adamax:u.OptimizerConstructors.adamax,adam:u.OptimizerConstructors.adam}},{"./optimizers/adadelta_optimizer":78,"./optimizers/adagrad_optimizer":79,"./optimizers/adam_optimizer":80,"./optimizers/adamax_optimizer":81,"./optimizers/momentum_optimizer":82,"./optimizers/optimizer_constructors":84,"./optimizers/rmsprop_optimizer":85,"./optimizers/sgd_optimizer":86}],94:[function(e,t,n){"use strict";function r(e,t){return s[e][t]}Object.defineProperty(n,"__esModule",{value:!0});!function(e){e.float32="float32",e.int32="int32",e.bool="bool"}(n.DType||(n.DType={}));!function(e){e.R0="R0",e.R1="R1",e.R2="R2",e.R3="R3",e.R4="R4"}(n.Rank||(n.Rank={}));var i;!function(e){e.float32="float32",e.int32="int32",e.bool="int32"}(i||(i={}));var a;!function(e){e.float32="float32",e.int32="int32",e.bool="bool"}(a||(a={}));var o;!function(e){e.float32="float32",e.int32="float32",e.bool="float32"}(o||(o={}));var s={float32:o,int32:i,bool:a};n.upcastType=r,n.sumOutType=function(e){return r(e,"int32")}},{}],95:[function(e,t,n){"use strict";function r(e,t,n){a(e instanceof h.Tensor,"Argument '"+t+"' passed to '"+n+"' must be a Tensor, but got "+typeof e+".")}function i(e){for(var t=e.length,n=0,r=0;t>0;)r=Math.random()*t|0,n=e[--t],e[t]=e[r],e[r]=n}function a(e,t){if(!e)throw new Error(t)}function o(e,t){if(void 0===t&&(t=[]),Array.isArray(e))for(var n=0;n<e.length;++n)o(e[n],t);else t.push(e);return t}function s(e,t){if(e.length!==t.length)return!1;for(var n=0;n<e.length;n++)if(e[n]!==t[n])return!1;return!0}function u(e,t,n){e[decodeURIComponent(t)]=decodeURIComponent(n||"")}function l(e){return e instanceof Float32Array||e instanceof Int32Array||e instanceof Uint8Array}function c(e){if(null==e)return[];if(e instanceof h.Tensor)return[e];var t=[],n=e;if(!p(n))return[];for(var r in n){var i=o(n[r]).filter(function(e){return e instanceof h.Tensor});t.push.apply(t,i)}return t}function p(e){return Array.isArray(e)||"object"==typeof e}Object.defineProperty(n,"__esModule",{value:!0});var h=e("./tensor");n.assertArgumentsAreTensors=function(e,t){for(var n in e)!function(n){var i=e[n];Array.isArray(i)?i.forEach(function(e,i){r(e,n+"["+i+"]",t)}):r(i,n,t)}(n)},n.shuffle=i,n.clamp=function(e,t,n){return Math.max(e,Math.min(t,n))},n.randUniform=function(e,t){return Math.random()*(t-e)+e},n.distSquared=function(e,t){for(var n=0,r=0;r<e.length;r++){var i=Number(e[r])-Number(t[r]);n+=i*i}return n},n.assert=a,n.assertShapesMatch=function(e,t,n){void 0===n&&(n=""),a(s(e,t),n+" Shapes "+e+" and "+t+" must match")},n.assertTypesMatch=function(e,t){a(e.dtype===t.dtype," The dtypes of the first("+e.dtype+") and second("+t.dtype+") input must match")},n.flatten=o,n.inferShape=function(e){if(l(e))return[e.length];if(!Array.isArray(e))return[];for(var t=[];e instanceof Array;)t.push(e.length),e=e[0];return t},n.sizeFromShape=function(e){if(0===e.length)return 1;for(var t=e[0],n=1;n<e.length;n++)t*=e[n];return t},n.isScalarShape=function(e){return 0===e.length},n.arraysEqual=s,n.isInt=function(e){return e%1==0},n.tanh=function(e){if(null!=Math.tanh)return Math.tanh(e);if(e===1/0)return 1;if(e===-1/0)return-1;var t=Math.exp(2*e);return(t-1)/(t+1)},n.sizeToSquarishShape=function(e){for(var t=Math.floor(Math.sqrt(e));t>1;--t)if(e%t==0)return[t,e/t];return[1,e]},n.createShuffledIndices=function(e){for(var t=new Uint32Array(e),n=0;n<e;++n)t[n]=n;return i(t),t},n.rightPad=function(e,t){return t<=e.length?e:e+" ".repeat(t-e.length)},n.repeatedTry=function(e,t,n){return void 0===t&&(t=function(e){return 0}),new Promise(function(r,i){var a=0,o=function(){if(e())r();else{var s=t(++a);null!=n&&a>=n?i():setTimeout(o,s)}};setTimeout(o,0)})},n.getQueryParams=function(e){var t={};return e.replace(/[?&]([^=?&]+)(?:=([^&]*))?/g,function(e){for(var n=[],r=1;r<arguments.length;r++)n[r-1]=arguments[r];return u(t,n[0],n[1]),n.join("=")}),t},n.inferFromImplicitShape=function(e,t){for(var n=1,r=-1,i=0;i<e.length;++i)if(e[i]>0)n*=e[i];else if(-1===e[i]){if(-1!==r)throw Error("Shapes can only have 1 implicit size. Found - 1 at dim "+r+" and dim "+i);r=i}else if(e[i]<=0)throw Error("Shapes can not be <= 0. Found "+e[i]+" at dim "+i);if(-1===r){if(t>0&&t!==n)throw Error("Size("+t+") must match the product of shape "+e);return e}if(t%n!=0)throw Error("The implicit shape can't be a fractional number. Got "+t+" / "+n);var a=e.slice();return a[r]=t/n,a},n.squeezeShape=function(e,t){for(var n=[],r=[],i=0,a=0;a<e.length;++a){if(null!=t){if(t[i]===a&&e[a]>1)throw new Error("Can't squeeze axis "+a+" since its dim '"+e[a]+"' is not 1");(null==t[i]||t[i]>a)&&1===e[a]&&(n.push(e[a]),r.push(a)),t[i]<=a&&i++}e[a]>1&&(n.push(e[a]),r.push(a))}return{newShape:n,keptDims:r}},n.getTypedArrayFromDType=function(e,t){var n=null;if(null==e||"float32"===e)n=new Float32Array(t);else if("int32"===e)n=new Int32Array(t);else{if("bool"!==e)throw new Error("Unknown data type "+e);n=new Uint8Array(t)}return n},n.isTensorInList=function(e,t){for(var n=0;n<t.length;n++)if(t[n].id===e.id)return!0;return!1},n.checkForNaN=function(e,t,n){if("float32"===t)for(var r=0;r<e.length;r++)if(isNaN(e[r]))throw Error("The result of the '"+n+"' has NaNs.")},n.flattenNameArrayMap=function(e,t){var n=[];if(e instanceof h.Tensor)n.push(e);else for(var r=e,i=0;i<t.length;i++)n.push(r[t[i]]);return n},n.unflattenToNameArrayMap=function(e,t){if(e.length!==t.length)throw new Error("Cannot unflatten Tensor[], keys and arrays are not of same length.");for(var n={},r=0;r<e.length;r++)n[e[r]]=t[r];return n},n.hasEncodingLoss=function(e,t){return!("float32"===t||"int32"===t&&"float32"!==e||"bool"===t&&"bool"===e)},n.copyTypedArray=function(e,t){if(null==t||"float32"===t)return new Float32Array(e);if("int32"===t)return new Int32Array(e);if("bool"===t){for(var n=new Uint8Array(e.length),r=0;r<n.length;++r)0!==Math.round(e[r])&&(n[r]=1);return n}throw new Error("Unknown data type "+t)},n.isTypedArray=l,n.bytesPerElement=function(e){if("float32"===e||"int32"===e)return 4;if("bool"===e)return 1;throw new Error("Unknown dtype "+e)},n.isFunction=function(e){return!!(e&&e.constructor&&e.call&&e.apply)},n.extractTensorsFromContainer=function(e){return c(e)},n.extractTensorsFromAny=c},{"./tensor":89}],96:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});n.version="0.8.4"},{}],97:[function(e,t,n){"use strict";var r=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},i=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var a=e("./ops/ops"),o=e("./util"),s={float32:4,int32:4};n.loadWeights=function(e,t,n,u){return void 0===t&&(t=""),r(this,void 0,void 0,function(){var r,l,c,p,h,d,f,g,m,v,y;return i(this,function(i){switch(i.label){case 0:if(r=e.map(function(){return!1}),l={},c=null!=n?n.map(function(){return!1}):[],p=[],e.forEach(function(e,t){var i=0;e.weights.forEach(function(e){var a=s[e.dtype]*o.sizeFromShape(e.shape),u=function(){r[t]=!0,null==l[t]&&(l[t]=[]),l[t].push({manifestEntry:e,groupOffset:i,sizeBytes:a})};null!=n?n.forEach(function(t,n){t===e.name&&(u(),c[n]=!0)}):u(),p.push(e.name),i+=a})}),!c.every(function(e){return e}))throw h=n.filter(function(e,t){return!c[t]}),new Error("Could not find weights in manifest with names: "+h.join(", ")+". \nManifest JSON has weights with names: "+p.join(", ")+".");return d=r.reduce(function(e,t,n){return t&&e.push(n),e},[]),f=[],d.forEach(function(n){e[n].paths.forEach(function(e){var n=t+(t.endsWith("/")?"":"/")+e;f.push(fetch(n,u))})}),[4,Promise.all(f)];case 1:return g=i.sent(),[4,Promise.all(g.map(function(e){return e.arrayBuffer()}))];case 2:return m=i.sent(),v={},y=0,d.forEach(function(t){for(var n=e[t].paths.length,r=0,i=0;i<n;i++)r+=m[y+i].byteLength;for(var o=new ArrayBuffer(r),s=new Uint8Array(o),u=0,c=0;c<n;c++){var p=new Uint8Array(m[y+c]);s.set(p,u),u+=p.byteLength}l[t].forEach(function(e){var t,n=o.slice(e.groupOffset,e.groupOffset+e.sizeBytes);if("float32"===e.manifestEntry.dtype)t=new Float32Array(n);else{if("int32"!==e.manifestEntry.dtype)throw new Error("Weight "+e.manifestEntry.name+" has unknown dtype "+e.manifestEntry.dtype+".");t=new Int32Array(n)}var r=e.manifestEntry.name;if(null!=v[r])throw new Error("Duplicate weight with name "+r+". Please make sure weights names are unique in the manifest JSON.");v[r]=a.tensor(t,e.manifestEntry.shape,e.manifestEntry.dtype)}),y+=n}),[2,v]}})})}},{"./ops/ops":66,"./util":95}],98:[function(e,t,n){"use strict";function r(e,t){return void 0===t&&(t=1),g.elu(e,t)}function i(e){return g.selu(e)}function a(e){return g.relu(e)}function o(e){return g.minimum(f.scalar(6),g.relu(e))}function s(e){return e}function u(e){return g.sigmoid(e)}function l(e){return g.hardSigmoid(e)}function c(e){return g.softplus(e)}function p(e){return g.softsign(e)}function h(e){return g.tanh(e)}function d(e,t){return void 0===t&&(t=-1),g.softmax(e,t)}Object.defineProperty(n,"__esModule",{value:!0});var f=e("@tensorflow/tfjs-core"),g=e("./backend/tfjs_backend"),m=e("./errors");n.getActivation=function(e){if(null==e)return s;if("elu"===e.toLowerCase())return r;if("hardsigmoid"===e.toLowerCase())return l;if("linear"===e.toLowerCase())return s;if("relu"===e.toLowerCase())return a;if("relu6"===e.toLowerCase())return o;if("selu"===e.toLowerCase())return i;if("sigmoid"===e.toLowerCase())return u;if("softmax"===e.toLowerCase())return d;if("softplus"===e.toLowerCase())return c;if("softsign"===e.toLowerCase())return p;if("tanh"===e.toLowerCase())return h;throw new m.ValueError("Unsupported activation function "+e)},n.elu=r,n.selu=i,n.relu=a,n.relu6=o,n.linear=s,n.sigmoid=u,n.hardSigmoid=l,n.softplus=c,n.softsign=p,n.tanh=h,n.softmax=d,n.serializeActivation=function(e){return e.name}},{"./backend/tfjs_backend":100,"./errors":107,"@tensorflow/tfjs-core":8}],99:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=1e-7;n.epsilon=function(){return r},n.setEpsilon=function(e){r=e},n.imageDataFormat=function(){return"channelsLast"}},{}],100:[function(e,t,n){"use strict";function r(){for(var e in re)for(var t in re[e])re[e][t].dispose(),delete re[e][t]}function i(e,t){return void 0===t&&(t=ne),null==re[t][e]&&(re[t][e]=K.scalar(e,t),H.keep(re[t][e])),re[t][e]}function a(e){return e.shape}function o(e){return e.shape.length}function s(e,t){if(null==t)return t;var n=a(e);return Array.isArray(t)?t.map(function(e){return Z.pyNormalizeArrayIndex(n,e)}):Z.pyNormalizeArrayIndex(n,t)}function u(e,t){return e.reshape(t)}function l(e,t){return H.transpose(e,t)}function c(e,t){return H.reverse(e,t)}function p(e,t){void 0===t&&(t=-1);var n=a(e).slice();return t<0&&(t=n.length+t+1),n.splice(t,0,1),u(e,n)}function h(e){return u(e,[J.arrayProd(e.shape)])}function d(e,t,n){switch(e.rank){case 1:return H.slice1d(e,t,n);case 2:return H.slice2d(e,[t,0],[n,e.shape[1]]);case 3:return H.slice3d(e,[t,0,0],[n,e.shape[1],e.shape[2]]);case 4:return H.slice4d(e,[t,0,0,0],[n,e.shape[1],e.shape[2],e.shape[3]]);default:throw new Y.ValueError("sliceAlongFirstAxis() received an unsupported tensor rank: "+e.rank)}}function f(e,t,n,r,i){return void 0===i&&(i=.001),K.tidy(function(){var a=H.moments(e,r),o=a.mean,s=a.variance;return[F(e,o,s,n,t,i),o,s]})}function g(e,t,n,r,i){return void 0===i&&(i=.001),K.tidy(function(){for(var a=H.moments(e,r),s=a.mean,l=a.variance,c=[],p=0,h=J.range(0,o(e));p<h.length;p++){var d=h[p];-1!==r.indexOf(d)?c.push(1):c.push(e.shape[d])}var f=u(s,c),g=u(l,c),m=null==t?null:u(t,c),v=null==n?null:u(n,c);return[F(e,f,g,v,m,i),s,l]})}function m(e,t){switch(e.rank){case 1:return H.concat1d([e,t]);case 2:return H.concat2d([e,t],0);case 3:return H.concat3d([e,t],0);case 4:return H.concat4d([e,t],0);default:throw new Y.ValueError("concatAlongFirstAxis() received an unsupported tensor rank: "+e.rank)}}function v(e,t){if(Array.isArray(t)||(t=[t]),o(e)!==t.length)throw new Y.ValueError("The length of input n ("+t.length+") does not match the number of dimensions in input x ("+o(e)+")");return H.tile(e,t)}function y(e,t){return H.zeros(e)}function b(e,t,n){for(var r=[],i=0;i<e;++i)for(var a=0;a<e;++a)r.push(i===a?1:0);return K.tensor2d(r,[e,e])}function w(e){return H.neg(e)}function x(e,t){return H.sub(e,t)}function A(e,t){return H.mul(e,t)}function E(e,t){return H.div(e,t)}function _(e,t){return H.mul(e,t)}function T(e,t){return H.add(e,t)}function S(e,t,n,r,i){return H.randomUniform(e,t,n)}function C(e,t,n,r,i){return void 0===t&&(t=0),void 0===n&&(n=1),H.truncatedNormal(e,t,n)}function O(e,t,n,r,i){if(void 0===t&&(t=0),void 0===n&&(n=1),r===Q.DType.bool)throw new Y.NotImplementedError("randomNormal does not support dType bool.");var a=r===Q.DType.float32?"float32":"int32";return H.randomNormal(e,t,n,a,i)}function k(e){var t=K.zerosLike(e),n=K.onesLike(e);return K.where(D(e,t),t,K.where(P(e,K.zerosLike(e)),n,_(i(-1),n)))}function R(e,t,n){return H.sum(e,t,n)}function N(e){return H.mulStrict(e,e)}function I(e){return H.sqrt(e)}function z(e){return H.log(e)}function M(e,t,n){return H.clipByValue(e,t,n)}function D(e,t){return H.equal(e,t)}function P(e,t){return H.greater(e,t)}function L(e,t){return H.maximum(e,t)}function F(e,t,n,r,i,a){void 0===a&&(a=.001);var s;if(2===o(e))s=H.batchNormalization2d(e,t,n,a,i,r);else if(3===o(e))s=H.batchNormalization3d(e,t,n,a,i,r);else{if(4!==o(e))throw new Y.NotImplementedError("batchNormalization is not implememnted for array of rank "+o(e)+" yet");s=H.batchNormalization4d(e,t,n,a,i,r)}return s}function V(e,t,n){if(null==n&&(n=ee.imageDataFormat()),X.checkDataFormat(n),1!==o(t)&&o(t)!==o(e))throw new Y.ValueError("Unexpected bias dimensions: "+o(t)+"; expected it to be 1 or "+o(e));var r,i=t.shape;if(5===o(e))"channelsFirst"===n?r=1===i.length?e.add(t.reshape([1,i[0],1,1,1])):e.add(t.reshape([1,i[3],i[0],i[1],i[2]])):"channelsLast"===n&&(r=1===i.length?e.add(t.reshape([1,1,1,1,i[0]])):e.add(t.reshape([1].concat(i))));else if(4===o(e))"channelsFirst"===n?r=1===i.length?e.add(t.reshape([1,i[0],1,1])):e.add(t.reshape([1,i[2],i[0],i[1]])):"channelsLast"===n&&(r=1===i.length?e.add(t.reshape([1,1,1,i[0]])):e.add(t.reshape([1].concat(i))));else if(3===o(e))"channelsFirst"===n?r=1===i.length?e.add(t.reshape([1,i[0],1])):e.add(t.reshape([1,i[1],i[0]])):"channelsLast"===n&&(r=1===i.length?e.add(t.reshape([1,1,i[0]])):e.add(t.reshape([1].concat(i))));else{if(!(o(e)<3))throw new Y.ValueError("Unsupported input rank by biasAdd: "+o(e));r=e.add(t)}return r}function B(e,t){return X.checkDataFormat(t),"channelsFirst"===t?H.transpose(e,[0,2,3,1]):e}function U(e,t,n,r,i,a,o){if(void 0===r&&(r=1),void 0===i&&(i="valid"),void 0===o&&(o=1),null==a&&(a=ee.imageDataFormat()),X.checkDataFormat(a),3!==e.shape.length)throw new Y.ValueError("The input of a conv1dWithBias operation should be 3, but is "+e.shape.length+" instead.");if(3!==t.shape.length)throw new Y.ValueError("The kernel for a conv1dWithBias operation should be 3, but is "+t.shape.length+" instead");if(null!=n&&1!==n.shape.length)throw new Y.ValueError("The bias for a conv1dWithBias operation should be 1, but is "+t.shape.length+" instead");if("channelsFirst"===a&&(e=l(e,[0,2,1])),"casual"===i)throw new Y.NotImplementedError("The support for CASUAL padding mode in conv1dWithBias is not implemented yet.");var s=H.conv1d(e,t,r,"same"===i?"same":"valid","NWC",o);return null!=n&&(s=V(s,n)),s}function j(e,t,n,r,i,a,s){if(void 0===r&&(r=[1,1]),void 0===i&&(i="valid"),null==a&&(a=ee.imageDataFormat()),X.checkDataFormat(a),3!==o(e)&&4!==o(e))throw new Y.ValueError("conv2dWithBias expects input to be of rank 3 or 4, but received "+o(e)+".");if(3!==o(t)&&4!==o(t))throw new Y.ValueError("conv2dWithBias expects kernel to be of rank 3 or 4, but received "+o(e)+".");var u=B(e,a);if("casual"===i)throw new Y.NotImplementedError("The support for CASUAL padding mode in conv1dWithBias is not implemented yet.");return u=H.conv2d(u,t,r,"same"===i?"same":"valid","NHWC",s),null!=n&&(u=V(u,n)),"channelsFirst"===a&&(u=H.transpose(u,[0,3,1,2])),u}function G(e,t){return void 0===t&&(t=-1),H.softmax(e,t)}function W(e,t,r){return void 0===r&&(r=!1),t=r?G(t):E(t,R(t,a(t).length-1,!0)),t=M(t,n.epsilon(),1-n.epsilon()),H.neg(H.sum(H.mul(e.toFloat(),H.log(t)),a(t).length-1))}function q(e,t){var n=H.maximum(t,H.zerosLike(t)),r=H.mul(t,e),a=H.log(H.add(i(1),H.exp(H.neg(H.abs(t)))));return H.add(H.sub(n,r),a)}Object.defineProperty(n,"__esModule",{value:!0});var H=e("@tensorflow/tfjs-core"),K=e("@tensorflow/tfjs-core"),X=e("../common"),Y=e("../errors"),Q=e("../types"),Z=e("../utils/generic_utils"),J=e("../utils/math_utils"),$=e("./common"),ee=e("./common"),te="webgl",ne=Q.DType.float32;n.disposeScalarCache=r,n.setBackend=function(e){H.setBackend(e),te=e,r()},n.getBackend=function(){return te},n.keep=function(e){return H.keep(e)};var re={float32:{},int32:{}};n.getScalar=i,n.epsilon=$.epsilon,n.isBackendSymbolic=function(){return!1},n.shape=a,n.intShape=function(e){return e.shape},n.ndim=o,n.dtype=function(e){return e instanceof K.Tensor?ne:e.dtype},n.normalizeAxis=s,n.countParams=function(e){var t=e.shape;return t.length>0?t.reduce(function(e,t){return e*t}):1},n.cast=function(e,t){return e.asType(t)},n.reshape=u,n.transpose=l,n.permuteDimensions=l,n.reverse=c,n.expandDims=p,n.squeeze=function(e,t){return H.squeeze(e,[t])},n.temporalPadding=function(e,t){if(3!==o(e))throw new Y.ValueError("temporalPadding expects input tensor to be 3-D, but received a "+o(e)+"-D tensor.");if(null==t&&(t=[1,1]),2!==t.length)throw new Y.ValueError("temporalPadding expects input padding pattern to be a length-2 array, but received a length-"+t.length+" array.");var n=[[0,0],t,[0,0]];return H.pad(e,n)},n.spatial2dPadding=function(e,t,n){if(4!==o(e))throw new Y.ValueError("temporalPadding expects input tensor to be 4-D, but received a "+o(e)+"-D tensor.");if(null==t&&(t=[[1,1],[1,1]]),2!==t.length||2!==t[0].length||2!==t[1].length)throw new Y.ValueError("spatial2dPadding expects `padding` to be an Array of two Arrays, each of which is an Array of two integers.");if(null==n&&(n=ee.imageDataFormat()),"channelsLast"!==n&&"channelsFirst"!==n)throw new Y.ValueError("Unknown data format: "+n+". Supported data formats are 'channelsLast' and 'channelsFirst.");var r;return r="channelsFirst"===n?[[0,0],[0,0],t[0],t[1]]:[[0,0],t[0],t[1],[0,0]],H.pad(e,r)},n.repeat=function(e,t){if(2!==e.shape.length)throw new Y.ValueError("repeat() expects a rank-2 tensor, but received a rank-"+e.shape.length+" tensor.");return v(p(e,1),[1,t,1])},n.flatten=h,n.batchFlatten=function(e){if(o(e)<=1)throw new Y.ValueError("batchFlatten requires a minimum rank of 2. Got rank: "+o(e)+".");return u(e,[e.shape[0],J.arrayProd(e.shape,1)])},n.sliceAlongFirstAxis=d,n.sliceAlongLastAxis=function(e,t,n){switch(e.rank){case 1:return H.slice1d(e,t,n);case 2:return H.slice2d(e,[0,t],[e.shape[0],n]);case 3:return H.slice3d(e,[0,0,t],[e.shape[0],e.shape[1],n]);case 4:return H.slice4d(e,[0,0,0,t],[e.shape[0],e.shape[1],e.shape[2],n]);default:throw new Y.ValueError("sliceAlongLastAxis() received an unsupported tensor rank: "+e.rank)}},n.normalizeBatchInTraining=function(e,t,n,r,i){return void 0===i&&(i=.001),K.util.arraysEqual(r.slice().sort(),J.range(0,o(e)-1))?f(e,t,n,r,i):g(e,t,n,r,i)},n.concatenate=function(e,t){void 0===t&&(t=-1);var n;return t<0&&(t=0!==(n=o(e[0]))?n:0),t===o(e[0])&&(t=-1),H.concat(e,t)},n.concatAlongFirstAxis=m,n.tile=v,n.variable=function(e,t,n,r){return new Q.LayerVariable(e,t,n,!0,r)},n.batchGetValue=function(e){return e.map(function(e){return e.read()})},n.batchSetValue=function(e){e.map(function(e){e[0].write(e[1])})},n.zeros=y,n.zerosVariable=function(e,t,n){return new Q.LayerVariable(y(e),t,n)},n.zerosLike=function(e,t,n){return new Q.LayerVariable(H.zerosLike(e),t,n)},n.ones=function(e,t){return H.ones(e)},n.onesVariable=function(e,t,n){var r=H.ones(e);return new Q.LayerVariable(r,t,n)},n.onesLike=function(e,t,n){var r=H.onesLike(e);return new Q.LayerVariable(r,t,n)},n.identity=function(e){return e.clone()},n.eye=b,n.eyeVariable=function(e,t,n){return new Q.LayerVariable(b(e,t),t,n)},n.neg=w,n.add=function(e,t){return H.add(e,t)},n.subtract=x,n.multiply=A,n.divide=E,n.scalarTimesArray=_,n.scalarPlusArray=T,n.randomUniform=S,n.randomUniformVariable=function(e,t,n,r,i,a){return void 0===a&&(a="randomUniform"),new Q.LayerVariable(S(e,t,n,r,i),r,a)},n.truncatedNormal=C,n.truncatedNormalVariable=function(e,t,n,r,i,a){return void 0===t&&(t=0),void 0===n&&(n=1),void 0===a&&(a="truncatedNormal"),new Q.LayerVariable(C(e,t,n,r,i),r,a)},n.randomNormal=O,n.randomNormalVariable=function(e,t,n,r,i,a){return void 0===t&&(t=0),void 0===n&&(n=1),void 0===a&&(a="randomNormal"),new Q.LayerVariable(O(e,t,n,r,i),r,a)},n.update=function(e,t){return e.write(t)},n.updateAdd=function(e,t){return e.write(H.add(e.read(),t))},n.updateSub=function(e,t){return e.write(H.sub(e.read(),t))},n.dot=function(e,t){if(2!==o(t))throw new Y.NotImplementedError("dot support for y other than rank 2 is not yet implemented: y shape = "+a);if(2===o(e))return H.matMul(e,t);if(3===o(e)){var n=e.shape[0],r=e.shape[1],i=e.shape[2];return e=e.reshape([n*r,i]),H.matMul(e,t).reshape([n,r,t.shape[1]])}throw new Y.NotImplementedError("dot support for x of rank "+o(e)+" is not yet implemented: x shape = "+a)},n.sign=k,n.qr=function(e){if(2!==e.shape.length)throw new Y.ValueError("qr() requires a 2D Tensor, but got a "+e.shape.length+"D Tensor.");if(e.shape[0]<e.shape[1])throw new Y.ValueError("qr() requires x.shape[0] >= x.shape[1], but got shape: ["+e.shape+"]");for(var t=e.shape[0],n=e.shape[1],r=b(t),i=e,a=K.tensor2d([[1]],[1,1]),o=0;o<n;++o){var s=i.slice([o,o],[t-o,1]),u=H.norm(s),l=i.slice([o,o],[1,1]),c=H.neg(k(l)),p=l.sub(A(c,u)),h=E(s,p),d=void 0;d=1===h.shape[0]?a:a.concat(h.slice([1,0],[h.shape[0]-1,h.shape[1]]),0);var f=H.neg(E(H.matMul(c,p),u)),g=i.slice([o,0],[t-o,n]),m=f.mul(d);i=0===o?g.sub(m.matMul(d.transpose().matMul(g))):i.slice([0,0],[o,n]).concat(g.sub(m.matMul(d.transpose().matMul(g))),0);var v=r.slice([0,o],[t,r.shape[1]-o]);r=0===o?v.sub(v.matMul(d).matMul(m.transpose())):r.slice([0,0],[t,o]).concat(v.sub(v.matMul(d).matMul(m.transpose())),1)}return[r,i]},n.oneHot=function(e,t){if(1!==o(e))throw new Error("Only 1D one-hot tensors are supported in the deeplearn backend, at present.");return e=e.toInt(),H.oneHot(e,t).toFloat()},n.mean=function(e,t,n){return t=s(e,t),H.mean(e,t,n)},n.argmax=function(e,t){return void 0===t&&(t=-1),H.argMax(e,t)},n.gather=function(e,t,n){return t=Array.isArray(t)?K.tensor1d(t,"int32"):t.toInt(),H.gather(e,t,n)},n.max=function(e,t,n){return H.max(e,t,n)},n.min=function(e,t,n){return H.min(e,t,n)},n.minimum=function(e,t){return H.minimum(e,t)},n.sum=R,n.abs=function(e){return H.abs(e)},n.square=N,n.sqrt=I,n.exp=function(e){return H.exp(e)},n.log=z,n.pow=function(e,t){if("number"==typeof t&&(t=K.scalar(Math.round(t),"int32")),"int32"!==t.dtype)throw new Y.NotImplementedError("Non-int32 dtype ("+t.dtype+") is not supported by pow() yet");return H.pow(e,t)},n.clip=M,n.equal=D,n.greater=P,n.greaterEqual=function(e,t){return H.greaterEqual(e,t)},n.maximum=L,n.sin=function(e){return H.sin(e.value())},n.cos=function(e){return H.cos(e.value())},n.batchNormalization=F,n.biasAdd=V,n.elu=function(e,t){if(void 0===t&&(t=1),1!==t)throw new Y.NotImplementedError("Support for alpha values other than 1 ("+t+") is not implemented yet.");return H.elu(e)},n.selu=function(e){return H.selu(e)},n.relu=function(e){return H.relu(e)},n.softplus=function(e){return H.log(H.add(i(1),H.exp(e)))},n.softsign=function(e){return H.div(e,H.add(i(1),H.abs(e)))},n.tanh=function(e){return H.tanh(e)},n.dropout=function(e,t,n,r){if(null!=n&&!K.util.arraysEqual(e.shape,n))throw new Y.NotImplementedError("Non-default noise shape is not implemented yet: "+JSON.stringify(n));if(null!=r)throw new Y.NotImplementedError("seed is not implemented for dropout yet.");var a=H.step(H.add(w(t),S(e.shape,0,1,Q.DType.float32)));return a=H.mul(E(i(1),x(i(1),t)),a),H.mul(e,a)},n.l2Normalize=function(e,t){return E(e,I(L(R(N(e),t,!0),_(K.scalar(n.epsilon()),H.onesLike(e)))))},n.conv1dWithBias=U,n.conv1d=function(e,t,n,r,i,a){return void 0===n&&(n=1),void 0===r&&(r="valid"),void 0===a&&(a=1),X.checkDataFormat(i),U(e,t,null,n,r,i,a)},n.conv2d=function(e,t,n,r,i,a){return void 0===n&&(n=[1,1]),void 0===r&&(r="valid"),X.checkDataFormat(i),j(e,t,null,n,r,i,a)},n.conv2dWithBias=j,n.depthwiseConv2d=function(e,t,n,r,i,a){void 0===n&&(n=[1,1]),void 0===r&&(r="valid"),null==i&&(i=ee.imageDataFormat()),X.checkDataFormat(i);var s=B(e,i);if(4!==o(e))throw new Y.ValueError("Input for depthwiseConv2d is required to be 4-D, but is instead "+o(e)+"-D");if(4!==o(t))throw new Y.ValueError("depthwiseKernel is required to be 4-D, but is instead "+o(t)+"-D");return s=H.depthwiseConv2d(s,t,n,"same"===r?"same":"valid","NHWC",a),"channelsFirst"===i&&(s=H.transpose(s,[0,3,1,2])),s},n.pool2d=function(e,t,n,r,i,a){X.checkDataFormat(i),X.checkPoolMode(a),X.checkPaddingMode(r),null==n&&(n=[1,1]),null==r&&(r="valid"),null==i&&(i=ee.imageDataFormat()),null==a&&(a="max"),e=B(e,i);var o,s="same"===r?"same":"valid";return o="max"===a?H.maxPool(e,t,n,s):H.avgPool(e,t,n,s),"channelsFirst"===i&&(o=H.transpose(o,[0,3,1,2])),o},n.nameScope=function(e,t){return X.nameScope(e,t)},n.floatx=function(){return Q.DType.float32};var ie={};n.getUid=function(e){return void 0===e&&(e=""),e in ie||(ie[e]=0),ie[e]+=1,e+ie[e].toString()},n.softmax=G,n.categoricalCrossentropy=W,n.sparseCategoricalCrossentropy=function(e,t,n){void 0===n&&(n=!1);var r=H.floor(h(e)).toInt(),i=a(t);return W(u(H.oneHot(r,i[i.length-1]),i),t,n)},n.binaryCrossentropy=function(e,t,r){void 0===r&&(r=!1);var i;return i=r?t:z(E(i=M(t,n.epsilon(),1-n.epsilon()),x(H.onesLike(i),i))),q(e,i)},n.sigmoidCrossEntropyWithLogits=q,n.sigmoid=function(e){return H.sigmoid(e)},n.hardSigmoid=function(e){return M(T(K.scalar(.5),_(K.scalar(.2),e)),0,1)},n.inTrainPhase=function(e,t,n){return void 0===n&&(n=!1),n?e():t()},n.rnn=function(e,t,n,r,i,a,o,s){void 0===r&&(r=!1),void 0===o&&(o=!1);var p=t.shape.length;if(p<3)throw new Y.ValueError("Input should be at least 3D, but is "+p+"D.");if(t=l(t,[1,0].concat(J.range(2,p))),null!=i)throw new Y.NotImplementedError("The rnn() function of the deeplearn.js backend does not support masking yet.");if(null!=a)throw new Y.NotImplementedError("The rnn() functoin of the deeplearn.js backend does not support constants yet.");o&&console.warn("Backend rnn(): the unroll = true option is not applicable to the imperative deeplearn.js backend."),r&&(t=c(t,0));for(var h,f,g=n,v=t.shape[0],y=0;y<v;++y){var b=d(t,y,1),w=e(b=u(b,b.shape.slice(1)),g);f=w[0],h=0===y?f.reshape([1].concat(f.shape)):m(h,f.reshape([1].concat(f.shape))),g=w[1]}return[f,l(h,[1,0].concat(J.range(2,h.shape.length))),g]},n.gradients=function(e,t){var n=t.map(function(e){return e.read()}),r=K.variableGrads(e,n);return t.map(function(e){return r.grads[e.name]})}},{"../common":102,"../errors":107,"../types":128,"../utils/generic_utils":130,"../utils/math_utils":131,"./common":99,"@tensorflow/tfjs-core":8}],101:[function(e,t,n){"use strict";function r(e){return a(this,void 0,void 0,function(){var t,n,r,i,a,s,u;return o(this,function(o){switch(o.label){case 0:if(null==e)return[2];t=[],n=[];for(r in e)"number"!=typeof(i=e[r])&&(a=i,t.push(a.data()),n.push(r));return[4,Promise.all(t)];case 1:for(s=o.sent(),u=0;u<s.length;++u)e[n[u]]=s[u][0];return[2]}})})}var i=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),a=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},o=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var s=e("@tensorflow/tfjs-core"),u=e("./backend/tfjs_backend"),l=e("./utils/generic_utils"),c=function(){function e(){this.validationData=null,this.model=null}return e.prototype.setParams=function(e){this.params=e},e.prototype.setModel=function(e){this.model=e},e.prototype.onEpochBegin=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e.prototype.onEpochEnd=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e.prototype.onBatchBegin=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e.prototype.onBatchEnd=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e.prototype.onTrainBegin=function(e){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e.prototype.onTrainEnd=function(e){return a(this,void 0,void 0,function(){return o(this,function(e){return[2]})})},e}();n.Callback=c;var p=function(){function e(e,t){void 0===t&&(t=10),null==e&&(e=[]),this.callbacks=e,this.queueLength=t}return e.prototype.append=function(e){this.callbacks.push(e)},e.prototype.setParams=function(e){for(var t=0,n=this.callbacks;t<n.length;t++)n[t].setParams(e)},e.prototype.setModel=function(e){for(var t=0,n=this.callbacks;t<n.length;t++)n[t].setModel(e)},e.prototype.onEpochBegin=function(e,t){return a(this,void 0,void 0,function(){var n,r,i;return o(this,function(a){switch(a.label){case 0:null==t&&(t={}),n=0,r=this.callbacks,a.label=1;case 1:return n<r.length?(i=r[n],[4,i.onEpochBegin(e,t)]):[3,4];case 2:a.sent(),a.label=3;case 3:return n++,[3,1];case 4:return[2]}})})},e.prototype.onEpochEnd=function(e,t){return a(this,void 0,void 0,function(){var n,r,i;return o(this,function(a){switch(a.label){case 0:null==t&&(t={}),n=0,r=this.callbacks,a.label=1;case 1:return n<r.length?(i=r[n],[4,i.onEpochEnd(e,t)]):[3,4];case 2:a.sent(),a.label=3;case 3:return n++,[3,1];case 4:return[2]}})})},e.prototype.onBatchBegin=function(e,t){return a(this,void 0,void 0,function(){var n,r,i;return o(this,function(a){switch(a.label){case 0:null==t&&(t={}),n=0,r=this.callbacks,a.label=1;case 1:return n<r.length?(i=r[n],[4,i.onBatchBegin(e,t)]):[3,4];case 2:a.sent(),a.label=3;case 3:return n++,[3,1];case 4:return[2]}})})},e.prototype.onBatchEnd=function(e,t){return a(this,void 0,void 0,function(){var n,r,i;return o(this,function(a){switch(a.label){case 0:null==t&&(t={}),n=0,r=this.callbacks,a.label=1;case 1:return n<r.length?(i=r[n],[4,i.onBatchEnd(e,t)]):[3,4];case 2:a.sent(),a.label=3;case 3:return n++,[3,1];case 4:return[2]}})})},e.prototype.onTrainBegin=function(e){return a(this,void 0,void 0,function(){var t,n,r;return o(this,function(i){switch(i.label){case 0:null==e&&(e={}),t=0,n=this.callbacks,i.label=1;case 1:return t<n.length?(r=n[t],[4,r.onTrainBegin(e)]):[3,4];case 2:i.sent(),i.label=3;case 3:return t++,[3,1];case 4:return[2]}})})},e.prototype.onTrainEnd=function(e){return a(this,void 0,void 0,function(){var t,n,r;return o(this,function(i){switch(i.label){case 0:null==e&&(e={}),t=0,n=this.callbacks,i.label=1;case 1:return t<n.length?(r=n[t],[4,r.onTrainEnd(e)]):[3,4];case 2:i.sent(),i.label=3;case 3:return t++,[3,1];case 4:return[2]}})})},e}();n.CallbackList=p;var h=function(e){function t(){return e.call(this)||this}return i(t,e),t.prototype.onEpochBegin=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(e){return this.seen=0,this.totals={},[2]})})},t.prototype.onBatchEnd=function(e,t){return a(this,void 0,void 0,function(){var e,n,r,i,a=this;return o(this,function(o){null==t&&(t={}),e=null==t.size?0:t.size,this.seen+=e,n=function(n){var i=t[n];"number"==typeof i?(r.totals.hasOwnProperty(n)||(r.totals[n]=0),r.totals[n]=r.totals[n]+i*e):(r.totals.hasOwnProperty(n)||(r.totals[n]=u.getScalar(0)),s.tidy(function(){a.totals[n]=u.scalarPlusArray(a.totals[n],u.multiply(i,u.getScalar(e))),u.keep(a.totals[n])}))},r=this;for(i in t)n(i);return[2]})})},t.prototype.onEpochEnd=function(e,t){return a(this,void 0,void 0,function(){var e,n,r,i,a,l=this;return o(this,function(o){if(null!=t)for(e=function(e){if(null==n.totals[e])return"continue";"number"==typeof n.totals[e]?t[e]=n.totals[e]/n.seen:s.tidy(function(){t[e]=u.scalarTimesArray(u.divide(u.getScalar(1),u.getScalar(l.seen)),l.totals[e]),u.keep(t[e])})},n=this,r=0,i=this.params.metrics;r<i.length;r++)a=i[r],e(a);return[2]})})},t}(c);n.BaseLogger=h,n.resolveScalarsInLogs=r,n.disposeTensorsInLogs=function(e){if(null!=e)for(var t in e){var n=e[t];"number"!=typeof n&&n.dispose()}};var d=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return i(t,e),t.prototype.onTrainBegin=function(e){return a(this,void 0,void 0,function(){return o(this,function(e){return this.epoch=[],this.history={},[2]})})},t.prototype.onEpochEnd=function(e,t){return a(this,void 0,void 0,function(){var n;return o(this,function(r){null==t&&(t={}),this.epoch.push(e);for(n in t)null==this.history[n]&&(this.history[n]=[]),this.history[n].push(t[n]);return[2]})})},t.prototype.syncData=function(){return a(this,void 0,void 0,function(){var e,t,n,r,i,a,s,u,l;return o(this,function(o){switch(o.label){case 0:e=[],t=[],n=[];for(r in this.history)for(i=this.history[r],a=0;a<i.length;++a)"number"!=typeof i[a]&&(s=i[a],e.push(s.data()),t.push(r),n.push(a));return[4,Promise.all(e)];case 1:for(u=o.sent(),l=0;l<u.length;++l)this.history[t[l]][n[l]].dispose(),this.history[t[l]][n[l]]=u[l][0];return[2]}})})},t}(c);n.History=d;var f=function(e){function t(t){var n=e.call(this)||this;return n.trainBegin=t.onTrainBegin,n.trainEnd=t.onTrainEnd,n.epochBegin=t.onEpochBegin,n.epochEnd=t.onEpochEnd,n.batchBegin=t.onBatchBegin,n.batchEnd=t.onBatchEnd,n}return i(t,e),t.prototype.onEpochBegin=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(n){switch(n.label){case 0:return null==this.epochBegin?[3,3]:[4,r(t)];case 1:return n.sent(),[4,this.epochBegin(e,t)];case 2:n.sent(),n.label=3;case 3:return[2]}})})},t.prototype.onEpochEnd=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(n){switch(n.label){case 0:return null==this.epochEnd?[3,3]:[4,r(t)];case 1:return n.sent(),[4,this.epochEnd(e,t)];case 2:n.sent(),n.label=3;case 3:return[2]}})})},t.prototype.onBatchBegin=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(n){switch(n.label){case 0:return null==this.batchBegin?[3,3]:[4,r(t)];case 1:return n.sent(),[4,this.batchBegin(e,t)];case 2:n.sent(),n.label=3;case 3:return[2]}})})},t.prototype.onBatchEnd=function(e,t){return a(this,void 0,void 0,function(){return o(this,function(n){switch(n.label){case 0:return null==this.batchEnd?[3,3]:[4,r(t)];case 1:return n.sent(),[4,this.batchEnd(e,t)];case 2:n.sent(),n.label=3;case 3:return[2]}})})},t.prototype.onTrainBegin=function(e){return a(this,void 0,void 0,function(){return o(this,function(t){switch(t.label){case 0:return null==this.trainBegin?[3,3]:[4,r(e)];case 1:return t.sent(),[4,this.trainBegin(e)];case 2:t.sent(),t.label=3;case 3:return[2]}})})},t.prototype.onTrainEnd=function(e){return a(this,void 0,void 0,function(){return o(this,function(t){switch(t.label){case 0:return null==this.trainEnd?[3,3]:[4,r(e)];case 1:return t.sent(),[4,this.trainEnd(e)];case 2:t.sent(),t.label=3;case 3:return[2]}})})},t}(c);n.CustomCallback=f,n.standardizeCallbacks=function(e){return null==e?null:e instanceof c?[e]:Array.isArray(e)&&e[0]instanceof c?e:l.toList(e).map(function(e){return new f(e)})}},{"./backend/tfjs_backend":100,"./utils/generic_utils":130,"@tensorflow/tfjs-core":8}],102:[function(e,t,n){"use strict";function r(){return 0===u.length?"":u.join(l)+l}function i(e){return!!e.match(c)}Object.defineProperty(n,"__esModule",{value:!0});var a=e("./errors"),o=e("./utils/generic_utils"),s=new Map;o.SerializableEnumRegistry.register("data_format",{channels_first:"channelsFirst",channels_last:"channelsLast"}),n.VALID_DATA_FORMAT_VALUES=["channelsFirst","channelsLast",void 0,null],n.checkDataFormat=function(e){if(null!=e&&n.VALID_DATA_FORMAT_VALUES.indexOf(e)<0)throw new a.ValueError(e+" is not a valid DataFormat.  Valid values as "+n.VALID_DATA_FORMAT_VALUES)},o.SerializableEnumRegistry.register("padding",{valid:"valid",same:"same",casual:"casual"}),n.VALID_PADDING_MODE_VALUES=["valid","same","casual",void 0,null],n.checkPaddingMode=function(e){if(null!=e&&n.VALID_PADDING_MODE_VALUES.indexOf(e)<0)throw new a.ValueError(e+" is not a valid PaddingMode.  Valid values as "+n.VALID_PADDING_MODE_VALUES)},n.VALID_POOL_MODE_VALUES=["max","avg",void 0,null],n.checkPoolMode=function(e){if(null!=e&&n.VALID_POOL_MODE_VALUES.indexOf(e)<0)throw new a.ValueError(e+" is not a valid PoolMode.  Valid values as "+n.VALID_POOL_MODE_VALUES)};var u=[],l="/";n.nameScope=function(e,t){u.push(e);try{var n=t();return u.pop(),n}catch(e){throw u.pop(),e}},n.getScopedTensorName=function(e){if(!i(e))throw new Error("Not a valid tensor name: '"+e+"'");return r()+e},n.getUniqueTensorName=function(e){if(!i(e))throw new Error("Not a valid tensor name: '"+e+"'");s.has(e)||s.set(e,0);var t=s.get(e);if(s.set(e,s.get(e)+1),t>0){var n=e+"_"+t;return s.set(n,1),n}return e};var c=new RegExp(/^[A-Za-z][A-Za-z0-9\._\/]*$/);n.isValidTensorName=i},{"./errors":107,"./utils/generic_utils":130}],103:[function(e,t,n){"use strict";function r(e,t){return u.sqrt(u.sum(u.square(e),t,!0))}function i(e,t){return void 0===t&&(t={}),c.deserializeKerasObject(e,c.ClassNameMap.getMap().pythonClassNameMap,t,"constraint")}var a=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),o=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var s=e("@tensorflow/tfjs-core"),u=e("./backend/tfjs_backend"),l=e("./types"),c=e("./utils/generic_utils"),p=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return a(t,e),t.prototype.getConfig=function(){return{}},t=o([s.doc({heading:"Constraints",subheading:"Classes",namespace:"constraints"})],t)}(l.Serializable);n.Constraint=p;var h=function(e){function t(t){var n=e.call(this)||this;return n.defaultMaxValue=2,n.defaultAxis=0,n.maxValue=null!=t.maxValue?t.maxValue:n.defaultMaxValue,n.axis=null!=t.axis?t.axis:n.defaultAxis,n}return a(t,e),t.prototype.apply=function(e){var t=r(e,this.axis),n=u.clip(t,0,this.maxValue);return u.multiply(e,u.divide(n,u.scalarPlusArray(u.getScalar(u.epsilon()),t)))},t.prototype.getClassName=function(){return"MaxNorm"},t.prototype.getConfig=function(){return{maxValue:this.maxValue,axis:this.axis}},t}(p);n.MaxNorm=h,c.ClassNameMap.register("MaxNorm",h);var d=function(e){function t(t){var n=e.call(this)||this;return n.defaultAxis=0,n.axis=null!=t.axis?t.axis:n.defaultAxis,n}return a(t,e),t.prototype.apply=function(e){return u.divide(e,u.scalarPlusArray(u.getScalar(u.epsilon()),r(e,this.axis)))},t.prototype.getClassName=function(){return"UnitNorm"},t.prototype.getConfig=function(){return{axis:this.axis}},t}(p);n.UnitNorm=d,c.ClassNameMap.register("UnitNorm",d);var f=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return a(t,e),t.prototype.apply=function(e){return u.relu(e)},t.prototype.getClassName=function(){return"NonNeg"},t}(p);n.NonNeg=f,c.ClassNameMap.register("NonNeg",f);var g=function(e){function t(t){var n=e.call(this)||this;return n.defaultMinValue=0,n.defaultMaxValue=1,n.defaultRate=1,n.defaultAxis=0,n.minValue=null!=t.minValue?t.minValue:n.defaultMinValue,n.maxValue=null!=t.maxValue?t.maxValue:n.defaultMaxValue,n.rate=null!=t.rate?t.rate:n.defaultRate,n.axis=null!=t.axis?t.axis:n.defaultAxis,n}return a(t,e),t.prototype.apply=function(e){var t=r(e,this.axis),n=u.add(u.scalarTimesArray(u.getScalar(this.rate),u.clip(t,this.minValue,this.maxValue)),u.scalarTimesArray(u.getScalar(1-this.rate),t));return u.multiply(e,u.divide(n,u.scalarPlusArray(u.getScalar(u.epsilon()),t)))},t.prototype.getClassName=function(){return"MinMaxNorm"},t.prototype.getConfig=function(){return{minValue:this.minValue,maxValue:this.maxValue,rate:this.rate,axis:this.axis}},t}(p);n.MinMaxNorm=g,c.ClassNameMap.register("MinMaxNorm",g),n.CONSTRAINT_IDENTIFIER_REGISTRY_SYMBOL_MAP={maxNorm:"MaxNorm",minMaxNorm:"MinMaxNorm",nonNeg:"NonNeg",unitNorm:"UnitNorm"},n.serializeConstraint=function(e){return c.serializeKerasObject(e)},n.deserializeConstraint=i,n.getConstraint=function(e){return null==e?null:"string"==typeof e?i({className:e in n.CONSTRAINT_IDENTIFIER_REGISTRY_SYMBOL_MAP?n.CONSTRAINT_IDENTIFIER_REGISTRY_SYMBOL_MAP[e]:e,config:{}}):e instanceof p?e:i(e)}},{"./backend/tfjs_backend":100,"./types":128,"./utils/generic_utils":130,"@tensorflow/tfjs-core":8}],104:[function(e,t,n){"use strict";function r(e,t){if(null!=e.dtype&&e.dtype!==t.dtype)throw new o.ValueError("The dtype of the feed ("+t.dtype+") is incompatible with that of the key '"+e.name+"' ("+e.dtype+").");if(null!=e.shape){if(e.shape.length!==t.shape.length)throw new o.ValueError("The rank of feed ("+t.shape.length+") does not match the rank of the key ("+e.shape.length+").");for(var n=0;n<e.shape.length;++n)if(null!=e.shape[n]&&e.shape[n]!==t.shape[n])throw new o.ValueError("The "+n+"-th dimension of the feed ("+t.shape[n]+") is incompatible with that of the key ("+e.shape[n]+").")}}function i(e,t,n){if(t.hasKey(e))return t.getValue(e);if(e.sourceLayer instanceof s.InputLayer)throw new o.ValueError("Missing a feed value for SymbolicTensor from InputLayer '"+s.InputLayer.name+"'");for(var r=[],u=0,l=e.inputs;u<l.length;u++){var c=i(l[u],t,n);r.push(c)}var p=e.sourceLayer.apply(r,n);Array.isArray(p)||(p=[p]);for(var h=a(e),d=Array.isArray(h)?h:[h],f=0;f<d.length;++f)t.add(d[f],p[f]);return 1===p.length?p[0]:p[e.outputTensorIndex]}function a(e){var t;if(1===e.sourceLayer.inboundNodes.length)t=e.sourceLayer.output;else{for(var n=null,r=0;r<e.sourceLayer.inboundNodes.length;++r)for(var i=0,a=e.sourceLayer.inboundNodes[r].outputTensors;i<a.length;i++)if(a[i].id===e.id){n=r;break}t=e.sourceLayer.getOutputAt(n)}return t}Object.defineProperty(n,"__esModule",{value:!0});var o=e("../errors"),s=e("./topology"),u=function(){function e(t){if(this.id2Value={},t instanceof e)for(var n in t.id2Value)this.id2Value[n]=t.id2Value[n];else{if(null==t)return;for(var r=0,i=t;r<i.length;r++){var a=i[r];this.add(a.key,a.value)}}}return e.prototype.add=function(e,t){if(r(e,t),null!=this.id2Value[e.id])throw new o.ValueError("Duplicate key: name="+e.name+", id="+e.id);return this.id2Value[e.id]=t,this},e.prototype.addFeed=function(e){this.add(e.key,e.value)},e.prototype.hasKey=function(e){return null!=this.id2Value[e.id]},e.prototype.getValue=function(e){if(null==this.id2Value[e.id])throw new o.ValueError("Nonexistent key: "+JSON.stringify(e));return this.id2Value[e.id]},e}();n.FeedDict=u,n.execute=function(e,t,n){for(var r=Array.isArray(e),a=r?e:[e],o=[],s=new u(t),l=0,c=a;l<c.length;l++){var p=c[l];o.push(i(p,s,n))}return r?o:o[0]}},{"../errors":107,"./topology":105}],105:[function(e,t,n){"use strict";function r(e){for(var t=[],n=0,r=e=v.toList(e);n<r.length;n++){var i=r[n];t.push(d.intShape(i))}return v.singletonOrArray(t)}function i(e){return m.DType.float32}function a(e,t,n){if((null==t||null!=n&&n>0)&&(t=e.sourceLayer,n=e.nodeIndex),0===t.inboundNodes.length)return[e];var r=t.inboundNodes[n];if(0===r.inboundLayers.length)return r.inputTensors;for(var i=[],o=0;o<r.inboundLayers.length;o++)for(var s=0,u=a(r.inputTensors[o],r.inboundLayers[o],r.nodeIndices[o]);s<u.length;s++){var l=u[s];-1===i.indexOf(l)&&i.push(l)}return i}function o(e,t,n){var r=v.stringToDType(e);return h.Tensor.make(t,{values:0===t.length?n:h.util.flatten(n)},r)}function s(e,t,n,r){if(!n.startsWith("2."))throw new f.ValueError("Unsupported Keras version in weights being loaded: "+n);return t}function u(e,t){for(var n={},r=0,i=0,a=t;i<a.length;i++)for(var o=0,s=a[i].weights;o<s.length;o++){var u=s[o];if(null!=n[u.name])throw new f.ValueError("Duplicate weight name: "+u.name);n[u.name]=u,r++}var l=[];for(var c in e)l.push([n[c],e[c]]),delete n[c];var p=[];for(var h in n)p.push(h);if(p.length>0)throw new f.ValueError(p.length+" of "+r+" weights are not set: "+p);d.batchSetValue(l)}function l(e,t,n){void 0===n&&(n=!1);for(var r=e.keras_version,i=e.backend,a=t.map(function(e){return e.name}),u={},l=0,c=t;l<c.length;l++)null!=(T=c[l]).name&&(null==u[T.name]&&(u[T.name]=[]),u[T.name].push(T));for(var p=e.weights,g=[],v=0;v<a.length;++v){var y=a[v],b=p[y];null==b&&(b=[]);for(var w=[],x=0;x<b.length;++x){var A=b[x];w.push(new m.LayerVariable(o(A.dtype,A.shape,A.value)))}for(var E=0,_=u[y];E<_.length;E++){var T=_[E],S=T.weights;if((w=s(T,w,r,i)).length!==S.length){if(!n)throw new f.ValueError("Layer #"+v+' (named "'+T.name+'") expects '+S.length+" weight(s), but the saved weights have "+w.length+" element(s).");console.warn("Skipping loading of weights of layer "+T.name+" due to mismatch in number of weights: ("+w.length+" vs "+S.length+").")}for(var C=0;C<w.length;++C)!n||h.util.arraysEqual(S[C].shape,w[C].shape)?g.push([S[C],w[C].read()]):console.warn("Skipping loading of weights for layer "+T.name+" due to mismatch in shape ("+S[C].shape+" vs "+w[C].shape+")")}}d.batchSetValue(g)}var c=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),p=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var h=e("@tensorflow/tfjs-core"),d=e("../backend/tfjs_backend"),f=e("../errors"),g=e("../layers/serialization"),m=e("../types"),v=e("../utils/generic_utils"),y=e("../utils/serialization_utils"),b=function(){return function(e){this.dtype=e.dtype,this.shape=e.shape,null!=e.shape?this.ndim=e.shape.length:this.ndim=e.ndim,this.maxNDim=e.maxNDim,this.minNDim=e.minNDim,this.axes=e.axes||{}}}();n.InputSpec=b;var w=0,x=function(){function e(e,t){this.callArgs=t,this.id=w++,this.outboundLayer=e.outboundLayer,this.inboundLayers=e.inboundLayers,this.nodeIndices=e.nodeIndices,this.tensorIndices=e.tensorIndices,this.inputTensors=e.inputTensors,this.outputTensors=e.outputTensors,this.inputMasks=e.inputMasks,this.outputMasks=e.outputMasks,this.inputShapes=e.inputShapes,this.outputShapes=e.outputShapes;for(var n=0,r=e.inboundLayers;n<r.length;n++){var i=r[n];null!=i&&i.outboundNodes.push(this)}e.outboundLayer.inboundNodes.push(this)}return e.prototype.getConfig=function(){for(var e=[],t=0,n=this.inboundLayers;t<n.length;t++){var r=n[t];null!=r?e.push(r.name):e.push(null)}return{outboundLayer:this.outboundLayer?this.outboundLayer.name:null,inboundLayers:e,nodeIndices:this.nodeIndices,tensorIndices:this.tensorIndices}},e}();n.Node=x;var A=0,E=function(e){function t(t){var n=e.call(this)||this;n._callHook=null,n._addedWeightNames=[],n._stateful=!1,n.id=A++,n.activityRegularizer=null,n.inputSpec=null,n.supportsMasking=!1,n._trainableWeights=[],n._nonTrainableWeights=[],n._losses=[],n._updates=[],n._built=!1,n.inboundNodes=[],n.outboundNodes=[];var r=t.name;if(!r){var i=n.getClassName();r=v.toSnakeCase(i)+"_"+d.getUid(i)}if(n.name=r,n.trainable=v.pyGetAttr(t,"trainable",!0),n.updatable=v.pyGetAttr(t,"updatable",!0),null!=t.inputShape||null!=t.batchInputShape){var a=void 0;if(null!=t.batchInputShape)a=t.batchInputShape;else if(null!=t.inputShape){var o=null;null!=t.batchSize&&(o=t.batchSize),a=[o].concat(t.inputShape)}n.batchInputShape=a;var s=t.dtype;null==s&&(s=t.inputDType),null==s&&(s=d.floatx()),n.dtype=s}return null!=t.weights?n.initialWeights=t.weights:n.initialWeights=null,n}return c(t,e),t.nodeKey=function(e,t){return e.name+"_ib-"+t.toString()},t.prototype.getNodeAtIndex=function(e,t){if(0===this.inboundNodes.length)throw new f.RuntimeError("The layer has never been called and thus has no defined "+t+".");if(this.inboundNodes.length<=e)throw new f.ValueError("Asked to get "+t+" at node "+e+", but the layer has only "+this.inboundNodes.length+" inbound nodes.");return this.inboundNodes[e]},t.prototype.getInputAt=function(e){return v.singletonOrArray(this.getNodeAtIndex(e,"input").inputTensors)},t.prototype.getOutputAt=function(e){return v.singletonOrArray(this.getNodeAtIndex(e,"output").outputTensors)},Object.defineProperty(t.prototype,"input",{get:function(){if(this.inboundNodes.length>1)throw new f.AttributeError("Layer "+this.name+' has multiple inbound nodes, hence the notion of "layer input" is ill-defined. Use `getInputAt(nodeIndex)` instead.');if(0===this.inboundNodes.length)throw new f.AttributeError("Layer "+this.name+" is not connected, no input to return.");return v.singletonOrArray(this.getNodeAtIndex(0,"input").inputTensors)},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"output",{get:function(){if(0===this.inboundNodes.length)throw new f.AttributeError("Layer "+this.name+" has no inbound nodes.");if(this.inboundNodes.length>1)throw new f.AttributeError("Layer "+this.name+' has multiple inbound nodes, hence the notion of "layer output" is ill-defined. Use `getOutputAt(nodeIndex)` instead.');return v.singletonOrArray(this.getNodeAtIndex(0,"output").outputTensors)},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"losses",{get:function(){return this._losses},enumerable:!0,configurable:!0}),t.prototype.calculateLosses=function(){return this.losses.map(function(e){return e()})},Object.defineProperty(t.prototype,"updates",{get:function(){return this._updates},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"built",{get:function(){return this._built},set:function(e){this._built=e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"trainableWeights",{get:function(){return this.trainable?this._trainableWeights:[]},set:function(e){this._trainableWeights=e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){return this.trainable?this._nonTrainableWeights:this._trainableWeights.concat(this._nonTrainableWeights)},set:function(e){this._nonTrainableWeights=e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"weights",{get:function(){return this.trainableWeights.concat(this.nonTrainableWeights)},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"stateful",{get:function(){return this._stateful},enumerable:!0,configurable:!0}),t.prototype.assertInputCompatibility=function(e){if(e=v.toList(e),null!=this.inputSpec&&0!==this.inputSpec.length){var t=v.toList(this.inputSpec);if(e.length!==t.length)throw new f.ValueError("Layer "+this.name+" expects "+t.length+" inputs, but it received "+e.length+" input tensors. Input received: "+e);for(var n=0;n<e.length;n++){var r=e[n],i=t[n];if(null!=i){var a=d.ndim(r);if(null!=i.ndim&&a!==i.ndim)throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+": expected ndim="+i.ndim+", found ndim="+a);if(null!=i.maxNDim&&a>i.maxNDim)throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+": expected max_ndim="+i.maxNDim+", found ndim="+a);if(null!=i.minNDim&&a<i.minNDim)throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+": expected min_ndim="+i.minNDim+", found ndim="+a+".");if(null!=i.dtype&&d.dtype(r)!==i.dtype){var o=d.dtype(r);throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+" : expected dtype="+i.dtype+", found dtype="+o+".")}if(i.axes){p=d.intShape(r);for(var s in i.axes){var u=Number(s),l=i.axes[s],c=u>=0?p[u]:p[p.length+u];if(null!=l&&-1===[l,null].indexOf(c))throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+": expected axis "+u+" of input shape to have value "+l+" but got shape "+p+".")}}if(null!=i.shape)for(var p=d.intShape(r),h=0;h<i.shape.length;++h){var g=i.shape[h],m=p[h];if(null!=g&&null!=m&&g!==m)throw new f.ValueError("Input "+n+" is incompatible with layer "+this.name+": expected shape="+i.shape+", found shape=${xShape}.")}}}}},t.prototype.call=function(e,t){return e},t.prototype.invokeCallHook=function(e,t){null!=this._callHook&&this._callHook(e,t)},t.prototype.setCallHook=function(e){this._callHook=e},t.prototype.clearCallHook=function(){this._callHook=null},t.prototype.apply=function(e,t){var n=this;t=t||{};for(var a=v.toList(e),o=!0,s=0,u=a;s<u.length;s++)if(!((h=u[s])instanceof m.SymbolicTensor)){o=!1;break}for(var l=!0,c=0,p=a;c<p.length;c++){var h=p[c];if(h instanceof m.SymbolicTensor){l=!1;break}}if(o===l)throw new f.ValueError("Arguments to apply() must be all SymbolicTensors or all Tensors");return d.nameScope(this.name,function(){if(!n.built){n.assertInputCompatibility(e);for(var o=[],s=0,u=v.toList(e);s<u.length;s++){var c=u[s];o.push(d.intShape(c))}n.build(v.singletonOrArray(o)),n.built=!0,n.initialWeights&&n.setWeights(n.initialWeights)}if(n.assertInputCompatibility(e),l){for(var p=n.call(e,t),h=[],g=0,y=v.toList(p);g<y.length;g++){var b=y[g];-1!==a.indexOf(b)&&(b=d.identity(b)),h.push(b)}if(p=v.singletonOrArray(h),null!=n.activityRegularizer)throw new f.NotImplementedError("Layer invocation in the presence of activity regularizer(s) is not supported yet.");return p}var w=r(e),x=n.computeOutputShape(w),p=void 0,A=i();if(p=null!=x&&x.length>0&&Array.isArray(x[0])?x.map(function(r,i){return new m.SymbolicTensor(A,r,n,v.toList(e),t,n.name,i)}):new m.SymbolicTensor(A,x,n,v.toList(e),t,n.name),n.addInboundNode(e,p,null,null,w,x,t),null!=n.activityRegularizer)throw new f.NotImplementedError("Layer invocation in the presence of activity regularizer(s) is not supported yet.");return p})},t.prototype.build=function(e){this.built=!0},t.prototype.getWeights=function(){return d.batchGetValue(this.weights)},t.prototype.setWeights=function(e){var t=this.weights;if(t.length!==e.length)throw new f.ValueError('You called setWeights(weights) on layer "'+this.name+'" with a weight list of length '+e.length+", but the layer was expecting "+t.length+" weights. Provided weights: "+e+"...");if(0!==t.length){for(var n=[],r=d.batchGetValue(t),i=0;i<r.length;++i){var a=r[i],o=t[i],s=e[i];if(!h.util.arraysEqual(a.shape,s.shape))throw new f.ValueError("Layer weight shape "+a.shape+" not compatible with provided weight shape "+s.shape);n.push([o,s])}d.batchSetValue(n)}},t.prototype.addWeight=function(e,t,n,r,i,a,o){if(-1!==this._addedWeightNames.indexOf(e))throw new f.ValueError("Duplicate weight name "+e+" for layer "+this.name);this._addedWeightNames.push(e),null==n&&(n=d.floatx());var s=new m.LayerVariable(r.apply(t,n),n,e,a,o);return null!=i&&this.addLoss(function(){return i.apply(s.read())}),null==a&&(a=!0),a?this._trainableWeights.push(s):this._nonTrainableWeights.push(s),s},t.prototype.addLoss=function(e){if(!(null==e||Array.isArray(e)&&0===e.length)){e=v.toList(e),void 0!==this._losses&&null!==this._losses&&(t=this.losses).push.apply(t,e);var t}},t.prototype.computeOutputShape=function(e){return e},t.prototype.computeMask=function(e,t){var n=this;if(!this.supportsMasking){if(null!=t){if(!Array.isArray(t))throw new TypeError("Layer "+this.name+" does not support masking,but was passed an inputMask.");t.forEach(function(e){if(null!=e)throw new TypeError("Layer "+n.name+" does not support masking,but was passed an inputMask.")})}return null}return t},t.prototype.addInboundNode=function(e,t,n,r,i,a,o){void 0===o&&(o=null);var s=v.toList(e);t=v.toList(t),n=v.toList(n),r=v.toList(r),i=v.normalizeShapeList(i),a=v.normalizeShapeList(a);for(var u=[],l=[],c=[],p=0,h=s;p<h.length;p++){var d=h[p];u.push(d.sourceLayer),l.push(d.nodeIndex),c.push(d.tensorIndex)}new x({outboundLayer:this,inboundLayers:u,nodeIndices:l,tensorIndices:c,inputTensors:s,outputTensors:t,inputMasks:n,outputMasks:r,inputShapes:i,outputShapes:a},o);for(var f=0;f<t.length;f++)t[f].sourceLayer=this,t[f].nodeIndex=this.inboundNodes.length-1,t[f].tensorIndex=f},t.prototype.getConfig=function(){var e={name:this.name,trainable:this.trainable};return null!=this.batchInputShape&&(e.batchInputShape=this.batchInputShape),null!=this.dtype&&(e.dtype=this.dtype),e},t.fromConfig=function(e,t){return new e(t)},p([h.doc({heading:"Models",subheading:"Classes"})],t.prototype,"apply",null),t=p([h.doc({heading:"Layers",subheading:"Classes",namespace:"layers"})],t)}(m.Serializable);n.Layer=E;var _=function(e){function t(t){var n=e.call(this,{dtype:t.dtype,name:null!=t.name?t.name:d.getUid("input").toString()})||this;if(null==t.batchSize&&(t.batchSize=null),null==t.sparse&&(t.sparse=!1),n.trainable=!1,n.built=!0,n.sparse=t.sparse,null!=t.inputShape&&null!=t.batchInputShape)throw new f.ValueError("Only provide the inputShape OR batchInputShape argument to inputLayer, not both at the same time.");var r=t.batchInputShape;if(null==r){if(null==t.inputShape)throw new f.ValueError("An InputLayer should be passed either a `batchInputShape` or an `inputShape`.");r=[t.batchSize].concat(t.inputShape)}else if(null!=t.batchSize)throw new f.ValueError("Cannot specify batchSize if batchInputShape isspecified when creating an InputLayer.");var i=t.dtype||d.floatx();n.batchInputShape=r,n.dtype=i,n.inputSpec=[{shape:r}];var a=new m.SymbolicTensor(n.dtype,n.batchInputShape,n,[],{},n.name);return a.nodeIndex=0,a.tensorIndex=0,new x({outboundLayer:n,inboundLayers:[],nodeIndices:[],tensorIndices:[],inputTensors:[a],outputTensors:[a],inputMasks:[null],outputMasks:[null],inputShapes:[r],outputShapes:[r]}),n}return c(t,e),t.prototype.apply=function(e,t){throw new f.ValueError("Cannot pass any input to an InputLayer's apply() method. InputLayer name: "+this.name)},t.prototype.getClassName=function(){return"InputLayer"},t.prototype.getConfig=function(){return{batchInputShape:this.batchInputShape,dtype:this.dtype,sparse:this.sparse,name:this.name}},t}(E);n.InputLayer=_,v.ClassNameMap.register("InputLayer",_),n.Input=function(e){if(null==e.batchShape&&null==e.shape)throw new Error("Please provide to Input either a `shape` or a `batchShape` argument. Note that `shape` does not include the batch dimension.");if(null!=e.batchShape&&null!=e.shape)throw new f.ValueError("Please provide either a `shape` or `batchShape` argument to Input, but not both.");var t=e.batchShape;null!=e.shape&&null==t&&(t=[null].concat(e.shape));var n=e.dtype;return null==n&&(n=d.floatx()),new _({batchInputShape:t,name:e.name,dtype:n,sparse:e.sparse}).inboundNodes[0].outputTensors[0]};var T=function(e){function t(n){var r=e.call(this,{})||this;if(r.containerNodes=new Set,r.name=n.name,null==r.name){var i=r.getClassName().toLowerCase();r.name=d.getUid(i)}if(r.supportsMasking=!1,r.trainable=!0,r.updatable=!0,Array.isArray(n.inputs)?r.inputs=n.inputs.slice():r.inputs=[n.inputs],Array.isArray(n.outputs)?r.outputs=n.outputs.slice():r.outputs=[n.outputs],v.unique(r.inputs).length!==r.inputs.length)throw new f.ValueError("The list of inputs passed to the model is redundant. All inputs should only appear once. Found: "+r.inputs.map(function(e){return e.name}));v.unique(r.outputs).length!==r.outputs.length&&console.warn("The list of outputs passed to the model is redundant. All outputs should only appear once. Found: "+r.outputs.map(function(e){return e.name})),r.inputLayers=[],r.inputLayersNodeIndices=[],r.inputLayersTensorIndices=[],r.outputLayers=[],r.outputLayersNodeIndices=[],r.outputLayersTensorIndices=[],r.layers=[];for(var a=0,o=r.outputs;a<o.length;a++){var s=(ae=o[a]).sourceLayer,u=ae.nodeIndex,l=ae.tensorIndex;r.outputLayers.push(s),r.outputLayersNodeIndices.push(u),r.outputLayersTensorIndices.push(l)}for(var c=0,p=r.inputs;c<p.length;c++){var s=(ae=p[c]).sourceLayer,u=ae.nodeIndex,l=ae.tensorIndex;v.assert(0===u,"input layer has >1 nodes"),v.assert(0===l,"input layer has >1 tensors"),r.inputLayers.push(s),r.inputLayersNodeIndices.push(u),r.inputLayersTensorIndices.push(l)}r.inputNames=[],r.outputNames=[],r.feedInputShapes=[],r.feedInputNames=[],r.feedOutputNames=[];for(M=0;M<r.inputLayers.length;M++){if(!((s=r.inputLayers[M])instanceof _))throw new TypeError("Input layers to a Model must be InputLayer objects. Received inputs: "+n.inputs+". Input "+M+" (0-based) originates from layer type "+s.getClassName()+".");r.inputNames.push(s.name),r.feedInputShapes.push(s.batchInputShape),r.feedInputNames.push(s.name)}for(var h=0,g=r.outputLayers;h<g.length;h++){s=g[h];r.outputNames.push(s.name)}r.internalInputShapes=r.inputs.map(function(e){return e.shape}),r.internalOutputShapes=r.outputs.map(function(e){return e.shape});for(var m={},y={},b={},w={},A={},E=[],T=function(e,n,i,a,o,s){null!=a&&null!=o&&null!=s||(a=e.sourceLayer,o=e.nodeIndex,s=e.tensorIndex);var u=a.inboundNodes[o];if(-1!==i.indexOf(u))throw new f.RuntimeError("The tensor "+e.name+' at layer "'+a.name+'" is part of a cycle.');if(-1===n.indexOf(u)){r.containerNodes.add(t.nodeKey(a,o)),a.id in A||(A[a.id]=Object.keys(A).length),-1===i.indexOf(u)&&i.push(u);for(var l=u.inboundLayers.length,c=0;c<l;c++){var p=u.inputTensors[c],h=u.inboundLayers[c],d=u.nodeIndices[c],g=u.tensorIndices[c];T(p,n,i,h,d,g)}for(n.push(u);i.indexOf(u)>=0;)i.splice(i.indexOf(u),1);E.push(u)}},S=[],C=[],O=0,k=r.outputs;O<k.length;O++){ae=k[O];T(ae,S,C)}for(var R=0,N=E.slice().reverse();R<N.length;R++){y[(ee=N[R]).id]=ee,ee.id in m||(m[ee.id]=0);var I=m[ee.id],z=null==b[ee.outboundLayer.id]?0:b[ee.outboundLayer.id];I=Math.max(I,z),b[ee.outboundLayer.id]=I,w[ee.outboundLayer.id]=ee.outboundLayer,m[ee.id]=I;for(var M=0;M<ee.inboundLayers.length;M++){var D=ee.inboundLayers[M],u=ee.nodeIndices[M],P=D.inboundNodes[u],L=null==m[P.id]?0:m[P.id];m[P.id]=Math.max(I+1,L),y[P.id]=P}}var F={};for(var V in m)(I=m[V])in F||(F[I]=[]),F[I].push(y[V]);var B={};for(var U in b)(I=b[U])in B||(B[I]=[]),B[I].push(w[U]);var j=Object.keys(B).map(function(e){return parseInt(e,10)}).sort(v.reverseNumberCompare);r.layers=[];for(var G=0,W=j;G<W.length;G++){var q=B[I=W[G]];q.sort(function(e,t){var n=A[e.id],r=A[t.id];return n<r?-1:n>r?1:0});for(var H=0,K=q;H<K.length;H++){s=K[H];r.layers.push(s)}}r.layersByDepth=B,j=Object.keys(F).map(function(e){return parseInt(e,10)}).sort(v.reverseNumberCompare);for(var X=r.inputs.slice(),Y=[],Q=0,Z=j;Q<Z.length;Q++)for(var J=0,$=F[I=Z[Q]];J<$.length;J++){var ee=$[J];if(null!=(s=ee.outboundLayer)){for(var te=0,ne=ee.inputTensors;te<ne.length;te++){ae=ne[te];if(-1===X.indexOf(ae))throw new f.RuntimeError("Graph disconnected: cannot obtain value for tensor "+ae+' at layer "'+s.name+'". The following previous layers were accessed without issue: '+Y)}for(var re=0,ie=ee.outputTensors;re<ie.length;re++){var ae=ie[re];X.push(ae)}Y.push(s.name)}}r.nodesByDepth=F;for(var oe=r.layers.map(function(e){return e.name}),se=0,ue=oe;se<ue.length;se++)!function(e){var t=oe.filter(function(t){return t===e}).length;if(1!==t)throw new f.RuntimeError('The name "'+e+'" is used '+t+" times in the model. All layer names should be unique. Layer names: "+JSON.stringify(oe))}(ue[se]);return r.outboundNodes=[],r.inboundNodes=[],new x({outboundLayer:r,inboundLayers:[],nodeIndices:[],tensorIndices:[],inputTensors:r.inputs,outputTensors:r.outputs,inputMasks:r.inputs.map(function(e){return null}),outputMasks:r.outputs.map(function(e){return null}),inputShapes:r.inputs.map(function(e){return e.shape}),outputShapes:r.outputs.map(function(e){return e.shape})}),r.built=!0,r}return c(t,e),Object.defineProperty(t.prototype,"trainableWeights",{get:function(){if(this._trainableWeights.length>0)throw new f.ValueError("Container instance unexpectedly contains _trainableWeights.The trainable weights of a Container are a union of the trainable weights of its consituent Layers. Its own _trainableWeights must remain an empty Array.");if(!this.trainable)return[];for(var e=[],t=0,n=this.layers;t<n.length;t++){var r=n[t];e=e.concat(r.trainableWeights)}return e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){for(var e=[],t=0,n=this.layers;t<n.length;t++){o=n[t];e.push.apply(e,o.nonTrainableWeights)}if(!this.trainable){for(var r=[],i=0,a=this.layers;i<a.length;i++){var o=a[i];r.push.apply(r,o.trainableWeights)}return r.concat(e)}return e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"weights",{get:function(){return this.trainableWeights.concat(this.nonTrainableWeights)},enumerable:!0,configurable:!0}),t.prototype.loadWeights=function(e,t,n){void 0===t&&(t=!1),void 0===n&&(n=!1),n?u(e,this.layers):l(e,this.layers,t)},t.prototype.updatedConfig=function(){var e=this.getConfig();return{className:this.getClassName(),config:e,kerasVersion:"tfjs-layers pre-release",backend:"TensorFlow.js"}},t.prototype.toJSON=function(e){var t=this.updatedConfig();return JSON.stringify(y.convertTsToPythonic(t))},t.prototype.call=function(e,t){e=v.toList(e);var n;return n="mask"in t?v.toList(t.mask):v.pyListRepeat(null,e.length),this.runInternalGraph(e,n)[0]},t.prototype.computeMask=function(e,t){e=v.toList(e);var n;return n=null==t?v.pyListRepeat(null,e.length):v.toList(t),this.runInternalGraph(e,n)[1]},t.prototype.computeOutputShape=function(e){var t=v.normalizeShapeList(e);if(t.length!==this.inputLayers.length)throw new f.ValueError("Invalid inputShape argument "+e+": model has "+this.inputLayers.length+" tensor inputs.");for(var n={},r=0;r<t.length;r++){var i=this.inputLayers[r],a=t[r];n[S=i.name+"_0_0"]=a}var o=Object.keys(this.nodesByDepth).map(function(e){return parseInt(e,10)}).sort(v.reverseNumberCompare);if(o.length>1)for(var s=0,u=o;s<u.length;s++)for(var l=u[s],c=0,p=this.nodesByDepth[l];c<p.length;c++){var h=p[c],i=h.outboundLayer;if(-1===this.inputLayers.map(function(e){return e.id}).indexOf(i.id)){for(var d=[],g=0;g<h.inboundLayers.length;g++){var m=h.inboundLayers[g],y=h.nodeIndices[g],b=h.tensorIndices[g],w=n[S=m.name+"_"+y+"_"+b];d.push(w)}for(var x=i.computeOutputShape(v.singletonOrArray(d)),A=v.normalizeShapeList(x),E=i.inboundNodes.indexOf(h),g=0;g<A.length;g++)n[S=i.name+"_"+E+"_"+g]=A[g]}}for(var _=[],T=[],r=0;r<this.outputLayers.length;r++){var i=this.outputLayers[r],E=this.outputLayersNodeIndices[r],b=this.outputLayersTensorIndices[r],S=i.name+"_"+E+"_"+b;T.push(S)}for(r=0;r<T.length;r++){var C=T[r];v.assert(C in n),_.push(n[C])}return v.singletonOrArray(_)},t.prototype.runInternalGraph=function(e,t){null==t&&(t=v.pyListRepeat(null,e.length));for(var n={},r=0;r<this.inputs.length;++r){var i=this.inputs[r],a=e[r],o=t[r];n[i.id]=[a,o]}for(var s=0,u=Object.keys(this.nodesByDepth).map(function(e){return parseInt(e,10)}).sort(v.reverseNumberCompare);s<u.length;s++)for(var l=u[s],c=0,p=this.nodesByDepth[l];c<p.length;c++){for(var h=p[c],d=h.outboundLayer,g=h.inputTensors,m=h.outputTensors,y=new Array,b=0,w=g;b<w.length;b++)(i=w[b]).id in n&&y.push(n[i.id]);if(y.length===g.length){var x={},A=void 0,E=void 0,_=void 0,T=void 0;if(null!=h.callArgs&&(x=h.callArgs),1===y.length){var S=y[0],C=S[0],O=S[1];null==x.mask&&(x.mask=O),_=v.toList(d.call(C,x)),T=v.toList(d.computeMask(C,O)),A=[C],E=[O]}else A=y.map(function(e){return e[0]}),E=y.map(function(e){return e[1]}),null==x.mask&&(x.mask=E),_=v.toList(d.call(A,x)),T=v.toList(d.computeMask(A,E));if(d.activityRegularizer)throw new f.NotImplementedError("Model invocation with concrete Tensor value(s) in the presence of activity regularizer(s) is not supported yet.");for(r=0;r<m.length;++r){var i=m[r],a=_[r],o=T[r];n[i.id]=[a,o]}}}for(var k=[],R=[],N=[],I=0,z=this.outputs;I<z.length;I++){i=z[I];v.assert(i.id in n,"Could not compute output "+i.name+" : "+i.id);var M=n[i.id],D=M[0],o=M[1];N.push(D.shape),k.push(D),R.push(o)}return[k,R,N]},t.prototype.buildNodeConversionMap=function(e){for(var n,r={},i=0,a=this.layers;i<a.length;i++){var o=a[i];n=o instanceof t?1:0;for(var s=0;s<o.inboundNodes.length;s++){var u=t.nodeKey(o,s);u in this.containerNodes&&(r[u]=n,n+=1)}}return r},t.prototype.getLayer=function(e,t){if(null!=t){if(this.layers.length<=t)throw new f.ValueError("Was asked to retrieve layer at index "+t+", but model only has "+this.layers.length+" layer(s).");return this.layers[t]}if(null==e)throw new f.ValueError("Provide either a layer name or layer index");for(var n=0,r=this.layers;n<r.length;n++){var i=r[n];if(i.name===e)return i}throw new f.ValueError("No such layer: "+e)},t.prototype.calculateLosses=function(){var e=this;return h.tidy(function(){for(var n=[],r=0,i=e.layers;r<i.length;r++)for(var a=i[r],o=0;o<a.inboundNodes.length;++o){var s=t.nodeKey(a,o);e.containerNodes.has(s)&&n.push.apply(n,a.calculateLosses())}return n})},t.prototype.getConfig=function(){for(var e={name:this.name},n=this.buildNodeConversionMap(this.layers),r=[],i=0,a=this.layers;i<a.length;i++){for(var o=(b=a[i]).getClassName(),s=b.getConfig(),u=[],l=0;l<b.inboundNodes.length;l++){var c=b.inboundNodes[l],p=t.nodeKey(b,l),h={};if(this.containerNodes.has(p)&&(c.callArgs&&(-1===JSON.stringify(c.callArgs).indexOf("undefined")?h=c.callArgs:(console.warn("Layer "+b.name+" was passed non-serializable keyword arguments: "+c.callArgs+". They will not be included in the serialized model (and thus will be missing at deserialization time)."),h={})),c.inboundLayers.length>0)){for(var d=[],f=0;f<c.inboundLayers.length;f++){var g=c.inboundLayers[f],m=c.nodeIndices[f],v=c.tensorIndices[f];null!==(x=n[t.nodeKey(g,m)])&&void 0!==x||(x=0),d.push([g.name,x,v,h])}u.push(d)}}r.push({name:b.name,className:o,config:s,inboundNodes:u})}e.layers=r;for(var y=[],f=0;f<this.inputLayers.length;f++){var b=this.inputLayers[f],m=this.inputLayersNodeIndices[f],p=t.nodeKey(b,m);if(this.containerNodes.has(p)){null!==(x=n[p])&&void 0!==x||(x=0);v=this.inputLayersTensorIndices[f];y.push([b.name,x,v])}}e.inputLayers=y;for(var w=[],f=0;f<this.outputLayers.length;f++){var b=this.outputLayers[f],m=this.outputLayersNodeIndices[f],p=t.nodeKey(b,m);if(this.containerNodes.has(p)){var x=n[p];null!==x&&void 0!==x||(x=0);v=this.outputLayersTensorIndices[f];w.push([b.name,x,v])}}return e.outputLayers=w,e},t.fromConfig=function(e,t){function n(e,t){e.name in i?i[e.name].push(t):i[e.name]=[t]}for(var r={},i={},a=t.name,o=t.layers,s=0,u=o;s<u.length;s++)!function(e){var i=e.name,a=g.deserialize(e,null!=t.customObjects?t.customObjects:{});r[i]=a;for(var o=0,s=e.inboundNodes;o<s.length;o++){var u=s[o];if(!(u instanceof Array))throw new f.ValueError("Corrupted configuration, expected array for nodeData: "+u);n(a,u)}}(p=u[s]);for(;!v.isObjectEmpty(i);)for(var l=0,c=o;l<c.length;l++){var p=c[l];if((S=r[p.name]).name in i){for(var h=0,d=i[S.name];h<d.length;h++)!function(e,t){for(var i,a=[],o=0,s=t;o<s.length;o++){var u=s[o],l=u[0],c=u[1],p=u[2];if(3===u.length)i={};else{if(4!==u.length)throw new f.ValueError("Improperly formatted model config for layer "+JSON.stringify(e)+": "+JSON.stringify(u));i=u[3]}if(!(l in r))return void n(e,t);var h=r[l];if(h.inboundNodes.length<=c)return void n(e,t);var d=h.inboundNodes[c];a.push(d.outputTensors[p])}a.length>0&&e.apply(v.singletonOrArray(a),i)}(S,d[h]);delete i[S.name]}}for(var m=[],y=[],b=0,w=t.inputLayers;b<w.length;b++){var x=(p=w[b])[0],A=p[1],E=p[2];v.assert(x in r);C=(S=r[x]).inboundNodes[A].outputTensors;m.push(C[E])}for(var _=0,T=t.outputLayers;_<T.length;_++){var x=(p=T[_])[0],A=p[1],E=p[2];v.assert(x in r);var S=r[x],C=S.inboundNodes[A].outputTensors;y.push(C[E])}return new e({inputs:m,outputs:y,name:a})},Object.defineProperty(t.prototype,"stateful",{get:function(){if(this._stateful)throw new f.ValueError("Container instance unexpectedly has _stateful = true. The statefulness of a Container is determined by the Layers it contains. Its _stateful property must remain the default false.");for(var e=0,t=this.layers;e<t.length;e++)if(t[e].stateful)return!0;return!1},enumerable:!0,configurable:!0}),p([h.doc({heading:"Layers",subheading:"Classes",namespace:"layers",subclasses:["Model"]})],t.prototype,"getLayer",null),t}(E);n.Container=T,n.getSourceInputs=a,n.loadWeightsFromNamedTensorMap=u,n.loadWeightsFromJson=l},{"../backend/tfjs_backend":100,"../errors":107,"../layers/serialization":121,"../types":128,"../utils/generic_utils":130,"../utils/serialization_utils":132,"@tensorflow/tfjs-core":8}],106:[function(e,t,n){"use strict";function r(e){return e instanceof b.Tensor}function i(e){return Array.isArray(e)}function a(e){return!r(e)&&!i(e)}function o(e,t,n,r,o){if(void 0===r&&(r=!0),void 0===o&&(o=""),null==t||0===t.length){if(null!=e){var s=!1;if(i(e)&&e.length>0)s=!0;else if(a(e)){for(var u in e)if(e.hasOwnProperty(u)){s=!0;break}}else s=!0;if(s)throw new A.ValueError("Error when checking model "+o+" expected no data, but got "+e)}return[]}if(null==e)return t.map(function(e){return null});var l;if(a(e)){e=e,l=[];for(var c=0,p=t;c<p.length;c++){var h=p[c];if(null==e[h])throw new A.ValueError('No data provided for "'+h+'". Need data for each key in: '+t);l.push(e[h])}}else if(i(e)){if((e=e).length!==t.length)throw new A.ValueError("Error when checking model "+o+": the Array of Tensors that you are passing to your model is not the size the model expected. Expected to see "+t.length+" Tensor(s), but instead got the following list of Tensor(s): "+e);l=e}else{if(e=e,t.length>1)throw new A.ValueError("The model "+o+" expects "+t.length+" Tensor(s), but only received one Tensor. Found: Tensor with shape "+e.shape);l=[e]}for(d=0;d<t.length;++d)1===(f=l[d]).shape.length&&(l[d]=w.expandDims(f,1));if(null!=n)for(var d=0;d<t.length;++d)if(null!=n[d]){var f=l[d];if(f.shape.length!==n[d].length)throw new A.ValueError("Error when checking "+o+": expected "+t[d]+" to have "+n[d].length+" dimension(s). but got array with shape "+f.shape);for(var g=0;g<n[d].length;++g)if(0!==g||r){var m=f.shape[g],v=n[d][g];if(null!=v&&v>=0&&m!==v)throw new A.ValueError("Error when checking "+o+": expected "+t[d]+" to have shape ["+n[d]+"], but got array with shape ["+f.shape+"].")}}return l}function s(e,t,n){var r=S.unique(e.map(function(e){return e.shape[0]}));r.sort();var i=S.unique(t.map(function(e){return e.shape[0]}));if(i.sort(),r.length>1)throw new A.ValueError("All input Tensors (x) should have the same number of samples. Got array shapes: "+JSON.stringify(e.map(function(e){return e.shape})));if(i.length>1)throw new A.ValueError("All target Tensors (y) should have the same number of samples. Got array shapes: "+JSON.stringify(t.map(function(e){return e.shape})));if(r.length>0&&i.length>0&&!b.util.arraysEqual(r,i))throw new A.ValueError("Input Tensors should have the same number of samples as target Tensors. Found "+r[0]+" input sample(s) and "+i[0]+" target sample(s).")}function u(e,t,n){for(var r=[E.meanSquaredError,E.binaryCrossentropy,E.categoricalCrossentropy],i=0;i<e.length;++i){var a=e[i],o=t[i],s=n[i];if(null!=o){if(o===E.categoricalCrossentropy&&1===a.shape[a.shape.length-1])throw new A.ValueError("You are passing a target array of shape "+a.shape+" while using a loss 'categorical_crossentropy'. 'categorical_crossentropy'expects targets to be binary matrices (1s and 0s) of shape [samples, classes].");if(-1!==r.indexOf(o))for(var u=a.shape.slice(1),l=s.slice(1),c=0;c<u.length;++c){var p=u[c],h=l[c];if(null!=h&&p!==h)throw new A.ValueError("A target Tensor with shape "+a.shape+" was passed for an output of shape "+s+", while using a loss function that expects targets to have the same shape as the output.")}}}}function l(e,t){for(var n=[],r=0,i=null;r<e;)(i=r+t)>=e&&(i=e),n.push([r,i]),r=i;return n}function c(e,t,n){return null==e?[null]:Array.isArray(e)?e.map(function(e){return w.sliceAlongFirstAxis(e,t,n-t)}):w.sliceAlongFirstAxis(e,t,n-t)}function p(e,t){return null==e?null:Array.isArray(e)?e.map(function(e){return p(e,t)}):w.gather(e,"int32"===t.dtype?t:t.toInt())}function h(e,t,n,r,i){void 0===r&&(r=!0),void 0===i&&(i="");var a;if(Array.isArray(e)){if(e.length!==t.length)throw new A.ValueError("Error when checking model "+i+": the Array of Tensors that you are passing to your model is not the size the the model expected. Expected to see "+t.length+" Tensor(s), but instead got "+e.length+" Tensors(s).");a=e}else{if(t.length>1)throw new A.ValueError("The model expects "+t.length+" "+i+" Tensors, but only received one Tensor. Found: array with shape "+JSON.stringify(e.shape)+".");a=[e]}if(null!=n)for(var o=0;o<t.length;++o)if(null!=n[o]){var s=a[o];if(s.shape.length!==n[o].length)throw new A.ValueError("Error when checking "+i+": expected "+t[o]+" to have "+n[o].length+" dimension(s), but got array with shape "+JSON.stringify(s.shape));for(var u=0;u<n[o].length;++u)if(0!==u||r){var l=s.shape[u],c=n[o][u];if(null!=c&&c!==l)throw new A.ValueError("Error when checking "+i+": expected "+t[o]+" to have shape "+JSON.stringify(n[o])+" but got array with shape "+JSON.stringify(s.shape)+".")}}}function d(e,t){if(null==e||Array.isArray(e)&&0===e.length)return t.map(function(e){return[]});if(Array.isArray(e))return t.map(function(t){return e});if(null!=e){for(var n=[],r=0,i=t;r<i.length;r++){var a=i[r],o=e.hasOwnProperty(a)?e[a]:[];Array.isArray(o)||(o=[o]),n.push(o)}return n}throw new TypeError("Type of metrics argument not understood. Expected an Array or Object, found: "+e)}var f=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),g=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o},m=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},v=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var y=e("@tensorflow/tfjs-core"),b=e("@tensorflow/tfjs-core"),w=e("../backend/tfjs_backend"),x=e("../callbacks"),A=e("../errors"),E=e("../losses"),_=e("../metrics"),T=e("../optimizers"),S=e("../utils/generic_utils"),C=e("../utils/math_utils"),O=e("./executor"),k=e("./topology");n.isDataTensor=r,n.isDataArray=i,n.isDataDict=a,n.standardizeInputData=o,n.checkArrayLengths=s,n.makeBatches=l,n.sliceArraysByIndices=p;!function(e){e[e.SILENT=0]="SILENT",e[e.VERBOSE=1]="VERBOSE"}(n.ModelLoggingVerbosity||(n.ModelLoggingVerbosity={}));var R=function(e){function t(t){return e.call(this,t)||this}return f(t,e),t.prototype.getClassName=function(){return"Model"},t.prototype.compile=function(e){var t=this;if(null==e.loss&&(e.loss=[]),this.loss=e.loss,"string"==typeof e.optimizer)this.optimizer=T.getOptimizer(e.optimizer);else{if(!(e.optimizer instanceof b.Optimizer))throw new A.ValueError("User-defined optimizer must be an instance of tf.Optimizer.");this.optimizer=e.optimizer}var n=[];if(Array.isArray(e.loss)||"string"==typeof e.loss||"function"==typeof e.loss)if(Array.isArray(e.loss)){if(e.loss.length!==this.outputs.length)throw new A.ValueError("When passing an Array as loss, it should have one entry per model output. The model has "+this.outputs.length+" output(s), but you passed loss="+e.loss+".");var r=e.loss;n=r.map(function(e){return E.get(e)})}else{var i=E.get(e.loss);this.outputs.map(function(e){n.push(i)})}else{e.loss=e.loss;for(var a in e.loss)if(-1===this.outputNames.indexOf(a))throw new A.ValueError('Unknown entry in loss dictionary: "'+a+'". Only expect the following keys: '+this.outputNames);for(var o in this.outputNames)null==e.loss[o]&&console.warn('Output "'+o+'" is missing from loss dictionary. We assume this was done on purpose, and we will not be expecting data to be passed to '+o+" during training"),n.push(E.get(e.loss[o]))}this.lossFunctions=n,this.feedOutputNames=[],this.feedOutputShapes=[],this.feedLossFns=[];for(var s=0;s<this.outputs.length;++s){var u=this.internalOutputShapes[s],l=this.outputNames[s];this.feedOutputNames.push(l),this.feedOutputShapes.push(u),this.feedLossFns.push(this.lossFunctions[s])}var c=[];this.metrics=e.metrics,this.metricsNames=["loss"],this.metricsTensors=[],w.nameScope("loss",function(){for(var e=0;e<t.outputs.length;++e)if(-1===c.indexOf(e)){var n=t.lossFunctions[e];t.outputs.length>1&&(t.metricsTensors.push([n,e]),t.metricsNames.push(t.outputNames[e]+"_loss"))}});var p=d(e.metrics,this.outputNames),h=function(e,n,r){t.outputNames.length>1&&(n=t.outputNames[e]+"_"+n),t.metricsNames.push(n),t.metricsTensors.push([r,e])};w.nameScope("metric",function(){for(var e=0;e<t.outputs.length;++e)!function(e){if(-1!==c.indexOf(e))return"continue";!function(n){for(var r,i,a,o=0,s=p[e];o<s.length;o++)!function(n){if(-1!==["accuracy","acc","crossentropy","ce"].indexOf(n)){var o=t.internalOutputShapes[e];1===o[o.length-1]||t.lossFunctions[e]===E.binaryCrossentropy?-1!==["accuracy","acc"].indexOf(n)?i=_.binaryAccuracy:-1!==["crossentropy","ce"].indexOf(n)&&(i=_.binaryCrossentropy):t.lossFunctions[e]===E.sparseCategoricalCrossentropy?-1!==["accuracy","acc"].indexOf(n)?i=_.sparseCategoricalAccuracy:-1!==["crossentropy","ce"].indexOf(n)&&(i=_.sparseCategoricalCrossentropy):-1!==["accuracy","acc"].indexOf(n)?i=_.categoricalAccuracy:-1!==["crossentropy","ce"].indexOf(n)&&(i=_.categoricalCrossentropy);var s=void 0;-1!==["accuracy","acc"].indexOf(n)?s="acc":-1!==["crossentropy","ce"].indexOf(n)&&(s="ce"),a=i,r=""+s}else{var u=_.get(n);a=u,r=""+n}var l;w.nameScope(r,function(){l=a}),h(e,r,l)}(s[o])}()}(e)}),this.collectedTrainableWeights=this.trainableWeights},t.prototype.checkTrainableWeightsConsistency=function(){null!=this.collectedTrainableWeights&&this.trainableWeights.length!==this.collectedTrainableWeights.length&&console.warn("Discrepancy between trainableweights and collected trainable weights. Did you set `model.trainable` without calling `model.compile()` afterwards?")},t.prototype.evaluate=function(e,t,n){void 0===n&&(n={});var r=null==n.batchSize?32:n.batchSize,i=this.standardizeUserData(e,t,!0,r),a=i[0].concat(i[1]);this.makeTestFunction();var o=this.testFunction,s=this.testLoop(o,a,r,n.verbose,n.steps);return S.singletonOrArray(s)},t.prototype.checkNumSamples=function(e,t,n,r){void 0===r&&(r="steps");var i;if(null!=n){if(i=null,null!=t)throw new A.ValueError("If "+r+" is set, batchSize must be null or undefined.Got batchSize = "+t)}else{if(null==e)throw new A.ValueError("Either the input data should have a defined shape, or "+r+" shoud be specified.");i=Array.isArray(e)?e[0].shape[0]:e.shape[0]}return i},t.prototype.predictLoop=function(e,t,n){var r=this;void 0===t&&(t=32),void 0===n&&(n=!1);var i=this.checkNumSamples(e);if(n)throw new A.NotImplementedError("Verbose predictLoop() is not implemented yet.");for(var a=l(i,t),o=[],s=0;s<a.length;++s)!function(t){var n=y.tidy(function(){var n=a[t][0],i=a[t][1],o=c(e,n,i),s=[];if(Array.isArray(o))for(var u=0;u<o.length;++u)s.push({key:r.inputs[u],value:o[u]});else s.push({key:r.inputs[0],value:o});var l=new O.FeedDict(s);return O.execute(r.outputs,l)});if(0===t)for(var i=0,s=n;i<s.length;i++){var u=s[i];o.push(u)}else for(var l=0;l<n.length;++l)o[l]=w.concatAlongFirstAxis(o[l],n[l])}(s);return S.singletonOrArray(o)},t.prototype.predict=function(e,t){void 0===t&&(t={}),h(e,this.inputNames,this.feedInputShapes,!1);var n=null==t.batchSize?32:t.batchSize;return this.predictLoop(e,n)},t.prototype.predictOnBatch=function(e){return h(e,this.inputNames,this.feedInputShapes,!0),this.predictLoop(e,e.shape[0])},t.prototype.standardizeUserData=function(e,t,n,r){if(void 0===n&&(n=!0),null==this.optimizer)throw new A.RuntimeError("You must compile a model before training/testing. Use Model.compile(modelCompileConfig).");for(var i=[],a=0;a<this.feedOutputShapes.length;++a){var l=this.feedOutputShapes[a];this.feedLossFns[a]===E.sparseCategoricalCrossentropy?i.push(l.slice(0,l.length-1).concat([1])):i.push(l)}if(e=o(e,this.feedInputNames,this.feedInputShapes,!1,"input"),t=o(t,this.feedOutputNames,i,!1,"target"),s(e,t),u(t,this.feedLossFns,this.feedOutputShapes),this.stateful&&null!=r&&r>0&&e[0].shape[0]%r!=0)throw new A.ValueError("In a stateful network, you should only pass inputs with a number of samples that is divisible by the batch size "+r+". Found: "+e[0].shape[0]+" sample(s).");return[e,t,null]},t.prototype.fitLoop=function(e,t,n,r,i,a,o,s,u,c,h,d,f,g){return void 0===d&&(d=0),m(this,void 0,void 0,function(){var m,E,_,T,S,O,k=this;return v(this,function(R){switch(R.label){case 0:if(null==r&&(r=32),null==i&&(i=100),null==c&&(c=!0),null==d&&(d=0),m=!1,null!=s&&null!=u&&(m=!0),null!=g&&(m=!0,null==f))throw new A.ValueError("Can only use `validationSteps` when doing step-wise training, i.e., `stepsPerEpoch` must be set.");if(null!=(E=this.checkNumSamples(t,r,f,"steps_per_epoch"))&&(_=C.range(0,E)),this.history=new x.History,o=null==o?[new x.BaseLogger]:[new x.BaseLogger].concat(o),o=o.concat([this.history]),a>0)throw new A.NotImplementedError("Verbose mode is not implemented yet.");return(T=new x.CallbackList(o)).setModel(this),T.setParams({epochs:i,steps:f,verbose:a,doValidation:m,metrics:h}),[4,T.onTrainBegin()];case 1:R.sent(),S=function(i){var a,o,h,d,g;return v(this,function(S){switch(S.label){case 0:return[4,T.onEpochBegin(i)];case 1:if(S.sent(),a={},null==f)return[3,2];throw new A.NotImplementedError("stepsPerEpoch mode is not implemented yet.");case 2:if("batch"===c)throw new A.NotImplementedError("batch shuffling is not implemneted yet");c&&b.util.shuffle(_),o=b.tensor1d(_),h=l(E,r),d=function(i){var l;return v(this,function(c){switch(c.label){case 0:return l={},[4,T.onBatchBegin(i,l)];case 1:return c.sent(),y.tidy(function(){var c=h[i][0],d=h[i][1],f=w.sliceAlongFirstAxis(o,c,d-c);l.batch=i,l.size=d-c;for(var g=p(t,f),v=e(g),y=0;y<n.length;++y){var b=n[y],x=v[y];l[b]=x,w.keep(x)}if(i===h.length-1&&m)for(var A=k.testLoop(s,u,r),y=0;y<n.length;++y){var b=n[y],x=A[y];w.keep(x),a["val_"+b]=x}}),[4,T.onBatchEnd(i,l)];case 2:return c.sent(),x.disposeTensorsInLogs(l),[2]}})},g=0,S.label=3;case 3:return g<h.length?[5,d(g)]:[3,6];case 4:S.sent(),S.label=5;case 5:return++g,[3,3];case 6:o.dispose(),S.label=7;case 7:return[4,T.onEpochEnd(i,a)];case 8:return S.sent(),[2]}})},O=d,R.label=2;case 2:return O<i?[5,S(O)]:[3,5];case 3:R.sent(),R.label=4;case 4:return++O,[3,2];case 5:return[4,T.onTrainEnd()];case 6:return R.sent(),[4,this.history.syncData()];case 7:return R.sent(),[2,this.history]}})})},t.prototype.testLoop=function(e,t,n,r,i){void 0===r&&(r=0);var a=this.checkNumSamples(t,n,i,"steps"),o=[];if(1===r)throw new A.NotImplementedError("Verbose mode is not implemented yet.");if(null!=i)throw new A.NotImplementedError("steps mode in testLoop() is not implemented yet");for(var s=l(a,n),u=b.tensor1d(C.range(0,a)),c=0;c<s.length;++c){var h=s[c][0],d=s[c][1],f=e(p(t,w.sliceAlongFirstAxis(u,h,d-h)));if(0===c)for(m=0;m<f.length;++m)o.push(w.getScalar(0));for(m=0;m<f.length;++m){var g=f[m];o[m]=w.add(o[m],w.scalarTimesArray(w.getScalar(d-h),g))}}for(var m=0;m<o.length;++m)o[m]=w.divide(o[m],w.getScalar(a));return o},t.prototype.getDedupedMetricsNames=function(){for(var e=this.metricsNames,t=[],n=0;n<e.length;++n){var r=e[n],i=r;S.count(e,r)>1&&(i+="_"+S.count(e.slice(0,n),r)),t.push(i)}return t},t.prototype.makeTestFunction=function(){var e=this;this.testFunction=function(t){return y.tidy(function(){for(var n,r=[],i=t.slice(0,e.inputs.length),a=t.slice(e.inputs.length,e.inputs.length+e.outputs.length),o=[],s=0;s<e.inputs.length;++s)o.push({key:e.inputs[s],value:i[s]});for(var u=new O.FeedDict(o),l=O.execute(e.outputs,u),s=0;s<e.lossFunctions.length;++s){var c=e.lossFunctions[s],p=w.mean(c(a[s],l[s]));n=0===s?p:w.add(n,p),r.push(n)}for(s=0;s<e.metricsTensors.length;++s){var h=e.metricsTensors[s][0],d=e.metricsTensors[s][1],f=w.mean(h(a[d],l[d]));r.push(f)}return r})}},t.prototype.fit=function(e,t,n){return void 0===n&&(n={}),m(this,void 0,void 0,function(){var r,i,a,o,s,u,l,p,h,d,f,g,m,y,b,E,_,T=this;return v(this,function(v){if(r=null==n.batchSize?32:n.batchSize,i=this.standardizeUserData(e,t,!1,r),a=i[0],o=i[1],s=!1,null!=n.validationData&&n.validationData.length>0){if(s=!0,2!==n.validationData.length)throw 3===n.validationData.length?new A.NotImplementedError("validationData including sample weights is not supported yet."):new A.ValueError("When passing validation data, it must contain 2 (valX, valY) or 3 (valX, valY, valSampleWeight) items; "+n.validationData+" is invalid.");u=n.validationData[0],l=n.validationData[1],h=this.standardizeUserData(u,l,!0,r),u=h[0],l=h[1],p=u.concat(l)}else null!=n.validationSplit&&n.validationSplit>0&&n.validationSplit<1?(s=!0,d=Math.floor(a[0].shape[0]*(1-n.validationSplit)),f=a[0].shape[0],u=c(a,d,f),a=c(a,0,d),l=c(o,d,f),o=c(o,0,d),p=u.concat(l)):null!=n.validationSteps&&(s=!0);return g=a.concat(o),this.checkTrainableWeightsConsistency(),m=function(e){var t=[],n=[],r=e.slice(0,T.inputs.length),i=e.slice(T.inputs.length,T.inputs.length+T.outputs.length),a=[],o=T.collectedTrainableWeights.map(function(e){return e.read()});return[T.optimizer.minimize(function(){for(var e=[],o=0;o<T.inputs.length;++o)e.push({key:T.inputs[o],value:r[o]});for(var s,u=new O.FeedDict(e),l=O.execute(T.outputs,u,{training:!0}),o=0;o<T.lossFunctions.length;++o){var c=(0,T.lossFunctions[o])(i[o],l[o]);t.push(c);var p=w.mean(c);n.push(p),s=0===o?c:w.add(s,c)}for(o=0;o<T.metricsTensors.length;++o){var h=T.metricsTensors[o][0],d=T.metricsTensors[o][1],f=w.mean(h(i[d],l[d]));w.keep(f),a.push(f)}return s=w.mean(s),T.calculateLosses().forEach(function(e){s=w.add(s,e)}),s},!0,o)].concat(a)},y=this.getDedupedMetricsNames(),s?(this.makeTestFunction(),b=this.testFunction,E=y.slice().concat(y.map(function(e){return"val_"+e}))):(b=null,p=[],E=y.slice()),_=x.standardizeCallbacks(n.callbacks),[2,this.fitLoop(m,g,y,r,n.epochs,n.verbose,_,b,p,n.shuffle,E,null,null,null)]})})},g([b.doc({heading:"Models",subheading:"Classes",configParamIndices:[0]})],t.prototype,"compile",null),g([b.doc({heading:"Models",subheading:"Classes",configParamIndices:[2]})],t.prototype,"evaluate",null),g([b.doc({heading:"Models",subheading:"Classes",configParamIndices:[1]})],t.prototype,"predict",null),g([b.doc({heading:"Models",subheading:"Classes"})],t.prototype,"predictOnBatch",null),g([b.doc({heading:"Models",subheading:"Classes",configParamIndices:[2]})],t.prototype,"fit",null),t=g([b.doc({heading:"Models",subheading:"Classes"})],t)}(k.Container);n.Model=R,S.ClassNameMap.register("Model",R)},{"../backend/tfjs_backend":100,"../callbacks":101,"../errors":107,"../losses":123,"../metrics":124,"../optimizers":126,"../utils/generic_utils":130,"../utils/math_utils":131,"./executor":104,"./topology":105,"@tensorflow/tfjs-core":8}],107:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.AttributeError=i;var a=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.RuntimeError=a;var o=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.ValueError=o;var s=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.NotImplementedError=s;var u=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.AssertionError=u;var l=function(e){function t(n){var r=e.call(this,n)||this;return Object.setPrototypeOf(r,t.prototype),r}return r(t,e),t}(Error);n.IndexError=l},{}],108:[function(e,t,n){"use strict";var r=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("./constraints"),o=e("./engine/topology"),s=e("./engine/training"),u=e("./initializers"),l=e("./layers/advanced_activations"),c=e("./layers/convolutional"),p=e("./layers/convolutional_depthwise"),h=e("./layers/core"),d=e("./layers/embeddings"),f=e("./layers/merge"),g=e("./layers/normalization"),m=e("./layers/padding"),v=e("./layers/pooling"),y=e("./layers/recurrent"),b=e("./layers/wrappers"),w=e("./losses"),x=e("./metrics"),A=e("./models"),E=e("./regularizers"),_=function(){function e(){}return e.model=function(e){return new s.Model(e)},e.sequential=function(e){return new A.Sequential(e)},e.loadModel=function(e){return A.loadModelInternal(e)},e.input=function(e){return o.Input(e)},r([i.doc({heading:"Models",subheading:"Creation",configParamIndices:[0]})],e,"model",null),r([i.doc({heading:"Models",subheading:"Creation",configParamIndices:[0]})],e,"sequential",null),r([i.doc({heading:"Models",subheading:"Loading",useDocsFrom:"loadModelInternal"})],e,"loadModel",null),r([i.doc({heading:"Models",subheading:"Inputs",useDocsFrom:"Input",configParamIndices:[0]})],e,"input",null),e}();n.ModelExports=_;var T=function(){function e(){}return e.inputLayer=function(e){return new o.InputLayer(e)},e.elu=function(e){return new l.ELU(e)},e.leakyReLU=function(e){return new l.LeakyReLU(e)},e.softmax=function(e){return new l.Softmax(e)},e.thresholdedReLU=function(e){return new l.ThresholdedReLU(e)},e.conv1d=function(e){return new c.Conv1D(e)},e.conv2d=function(e){return new c.Conv2D(e)},e.conv2dTranspose=function(e){return new c.Conv2DTranspose(e)},e.separableConv2d=function(e){return new c.SeparableConv2D(e)},e.depthwiseConv2d=function(e){return new p.DepthwiseConv2D(e)},e.activation=function(e){return new h.Activation(e)},e.dense=function(e){return new h.Dense(e)},e.dropout=function(e){return new h.Dropout(e)},e.flatten=function(e){return new h.Flatten(e)},e.repeatVector=function(e){return new h.RepeatVector(e)},e.reshape=function(e){return new h.Reshape(e)},e.embedding=function(e){return new d.Embedding(e)},e.add=function(e){return new f.Add(e)},e.average=function(e){return new f.Average(e)},e.concatenate=function(e){return new f.Concatenate(e)},e.maximum=function(e){return new f.Maximum(e)},e.minimum=function(e){return new f.Minimum(e)},e.multiply=function(e){return new f.Multiply(e)},e.batchNormalization=function(e){return new g.BatchNormalization(e)},e.zeroPadding2d=function(e){return new m.ZeroPadding2D(e)},e.averagePooling1d=function(e){return new v.AveragePooling1D(e)},e.avgPool1d=function(t){return e.averagePooling1d(t)},e.avgPooling1d=function(t){return e.averagePooling1d(t)},e.averagePooling2d=function(e){return new v.AveragePooling2D(e)},e.avgPool2d=function(t){return e.averagePooling2d(t)},e.avgPooling2d=function(t){return e.averagePooling2d(t)},e.globalAveragePooling1d=function(e){return new v.GlobalAveragePooling1D(e)},e.globalAveragePooling2d=function(e){return new v.GlobalAveragePooling2D(e)},e.globalMaxPooling1d=function(e){return new v.GlobalMaxPooling1D(e)},e.globalMaxPooling2d=function(e){return new v.GlobalMaxPooling2D(e)},e.maxPooling1d=function(e){return new v.MaxPooling1D(e)},e.maxPooling2d=function(e){return new v.MaxPooling2D(e)},e.gru=function(e){return new y.GRU(e)},e.gruCell=function(e){return new y.GRUCell(e)},e.lstm=function(e){return new y.LSTM(e)},e.lstmCell=function(e){return new y.LSTMCell(e)},e.simpleRNN=function(e){return new y.SimpleRNN(e)},e.simpleRNNCell=function(e){return new y.SimpleRNNCell(e)},e.rnn=function(e){return new y.RNN(e)},e.stackedRNNCells=function(e){return new y.StackedRNNCells(e)},e.bidirectional=function(e){return new b.Bidirectional(e)},e.timeDistributed=function(e){return new b.TimeDistributed(e)},e.Layer=o.Layer,e.RNNCell=y.RNNCell,e.input=_.input,r([i.doc({heading:"Layers",subheading:"Inputs",namespace:"layers",useDocsFrom:"InputLayer",configParamIndices:[0]})],e,"inputLayer",null),r([i.doc({heading:"Layers",subheading:"Advanced Activation",namespace:"layers",useDocsFrom:"ELU",configParamIndices:[0]})],e,"elu",null),r([i.doc({heading:"Layers",subheading:"Advanced Activation",namespace:"layers",useDocsFrom:"LeakyReLU",configParamIndices:[0]})],e,"leakyReLU",null),r([i.doc({heading:"Layers",subheading:"Advanced Activation",namespace:"layers",useDocsFrom:"Softmax",configParamIndices:[0]})],e,"softmax",null),r([i.doc({heading:"Layers",subheading:"Advanced Activation",namespace:"layers",useDocsFrom:"ThresholdedReLU",configParamIndices:[0]})],e,"thresholdedReLU",null),r([i.doc({heading:"Layers",subheading:"Convolutional",namespace:"layers",useDocsFrom:"Conv1D",configParamIndices:[0]})],e,"conv1d",null),r([i.doc({heading:"Layers",subheading:"Convolutional",namespace:"layers",useDocsFrom:"Conv2D",configParamIndices:[0]})],e,"conv2d",null),r([i.doc({heading:"Layers",subheading:"Convolutional",namespace:"layers",useDocsFrom:"Conv2DTranspose",configParamIndices:[0]})],e,"conv2dTranspose",null),r([i.doc({heading:"Layers",subheading:"Convolutional",namespace:"layers",useDocsFrom:"SeparableConv2D",configParamIndices:[0]})],e,"separableConv2d",null),r([i.doc({heading:"Layers",subheading:"Convolutional",namespace:"layers",useDocsFrom:"DepthwiseConv2D",configParamIndices:[0]})],e,"depthwiseConv2d",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Activation",configParamIndices:[0]})],e,"activation",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Dense",configParamIndices:[0]})],e,"dense",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Dropout",configParamIndices:[0]})],e,"dropout",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Flatten",configParamIndices:[0]})],e,"flatten",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"RepeatVector",configParamIndices:[0]})],e,"repeatVector",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Reshape",configParamIndices:[0]})],e,"reshape",null),r([i.doc({heading:"Layers",subheading:"Basic",namespace:"layers",useDocsFrom:"Embedding",configParamIndices:[0]})],e,"embedding",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Add",configParamIndices:[0]})],e,"add",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Average",configParamIndices:[0]})],e,"average",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Concatenate",configParamIndices:[0]})],e,"concatenate",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Maximum",configParamIndices:[0]})],e,"maximum",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Minimum",configParamIndices:[0]})],e,"minimum",null),r([i.doc({heading:"Layers",subheading:"Merge",namespace:"layers",useDocsFrom:"Multiply",configParamIndices:[0]})],e,"multiply",null),r([i.doc({heading:"Layers",subheading:"Normalization",namespace:"layers",useDocsFrom:"BatchNormalization",configParamIndices:[0]})],e,"batchNormalization",null),r([i.doc({heading:"Layers",subheading:"Padding",namespace:"layers",useDocsFrom:"ZeroPadding2D",configParamIndices:[0]})],e,"zeroPadding2d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"AveragePooling1D",configParamIndices:[0]})],e,"averagePooling1d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"AveragePooling2D",configParamIndices:[0]})],e,"averagePooling2d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"GlobalAveragePooling1D",configParamIndices:[0]})],e,"globalAveragePooling1d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"GlobalAveragePooling2D",configParamIndices:[0]})],e,"globalAveragePooling2d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"GlobalMaxPooling1D",configParamIndices:[0]})],e,"globalMaxPooling1d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"GlobalMaxPooling2D",configParamIndices:[0]})],e,"globalMaxPooling2d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"MaxPooling1D",configParamIndices:[0]})],e,"maxPooling1d",null),r([i.doc({heading:"Layers",subheading:"Pooling",namespace:"layers",useDocsFrom:"MaxPooling2D",configParamIndices:[0]})],e,"maxPooling2d",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"GRU",configParamIndices:[0]})],e,"gru",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"GRUCell",configParamIndices:[0]})],e,"gruCell",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"LSTM",configParamIndices:[0]})],e,"lstm",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"LSTMCell",configParamIndices:[0]})],e,"lstmCell",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"SimpleRNN",configParamIndices:[0]})],e,"simpleRNN",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"SimpleRNNCell",configParamIndices:[0]})],e,"simpleRNNCell",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"RNN",configParamIndices:[0]})],e,"rnn",null),r([i.doc({heading:"Layers",subheading:"Recurrent",namespace:"layers",useDocsFrom:"RNN",configParamIndices:[0]})],e,"stackedRNNCells",null),r([i.doc({heading:"Layers",subheading:"Wrapper",namespace:"layers",useDocsFrom:"Bidirectional",configParamIndices:[0]})],e,"bidirectional",null),r([i.doc({heading:"Layers",subheading:"Wrapper",namespace:"layers",useDocsFrom:"TimeDistributed",configParamIndices:[0]})],e,"timeDistributed",null),e}();n.LayerExports=T;var S=function(){function e(){}return e.maxNorm=function(e){return new a.MaxNorm(e)},e.unitNorm=function(e){return new a.UnitNorm(e)},e.nonNeg=function(){return new a.NonNeg},e.minMaxNorm=function(e){return new a.MinMaxNorm(e)},r([i.doc({heading:"Constraints",namespace:"constraints",useDocsFrom:"MaxNorm",configParamIndices:[0]})],e,"maxNorm",null),r([i.doc({heading:"Constraints",namespace:"constraints",useDocsFrom:"UnitNorm",configParamIndices:[0]})],e,"unitNorm",null),r([i.doc({heading:"Constraints",namespace:"constraints",useDocsFrom:"NonNeg"})],e,"nonNeg",null),r([i.doc({heading:"Constraints",namespace:"constraints",useDocsFrom:"MinMaxNormConfig",configParamIndices:[0]})],e,"minMaxNorm",null),e}();n.ConstraintExports=S;var C=function(){function e(){}return e.zeros=function(){return new u.Zeros},e.ones=function(){return new u.Ones},e.constant=function(e){return new u.Constant(e)},e.randomUniform=function(e){return new u.RandomUniform(e)},e.randomNormal=function(e){return new u.RandomNormal(e)},e.truncatedNormal=function(e){return new u.TruncatedNormal(e)},e.identity=function(e){return new u.Identity(e)},e.varianceScaling=function(e){return new u.VarianceScaling(e)},e.glorotUniform=function(e){return new u.GlorotUniform(e)},e.glorotNormal=function(e){return new u.GlorotNormal(e)},e.heNormal=function(e){return new u.HeNormal(e)},e.leCunNormal=function(e){return new u.LeCunNormal(e)},e.orthogonal=function(e){return new u.Orthogonal(e)},r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"Zeros"})],e,"zeros",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"Ones"})],e,"ones",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"Constant",configParamIndices:[0]})],e,"constant",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"RandomUniform",configParamIndices:[0]})],e,"randomUniform",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"RandomNormal",configParamIndices:[0]})],e,"randomNormal",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"TruncatedNormal",configParamIndices:[0]})],e,"truncatedNormal",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"Identity",configParamIndices:[0]})],e,"identity",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"VarianceScaling",configParamIndices:[0]})],e,"varianceScaling",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"GlorotUniform",configParamIndices:[0]})],e,"glorotUniform",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"GlorotNormal",configParamIndices:[0]})],e,"glorotNormal",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"HeNormal",configParamIndices:[0]})],e,"heNormal",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"LeCunNormal",configParamIndices:[0]})],e,"leCunNormal",null),r([i.doc({heading:"Initializers",namespace:"initializers",useDocsFrom:"Orthogonal",configParamIndices:[0]})],e,"orthogonal",null),e}();n.InitializerExports=C;var O=function(){function e(){}return e.binaryAccuracy=function(e,t){return x.binaryAccuracy(e,t)},e.binaryCrossentropy=function(e,t){return x.binaryCrossentropy(e,t)},e.categoricalAccuracy=function(e,t){return x.categoricalAccuracy(e,t)},e.categoricalCrossentropy=function(e,t){return w.categoricalCrossentropy(e,t)},e.cosineProximity=function(e,t){return w.cosineProximity(e,t)},e.prototype.meanAbsoluteError=function(e,t){return w.meanAbsoluteError(e,t)},e.prototype.meanAbsolutePercentageError=function(e,t){return w.meanAbsolutePercentageError(e,t)},e.prototype.MAPE=function(e,t){return w.meanAbsolutePercentageError(e,t)},e.prototype.mape=function(e,t){return w.meanAbsolutePercentageError(e,t)},e.meanSquaredError=function(e,t){return w.meanSquaredError(e,t)},e.MSE=function(e,t){return w.meanSquaredError(e,t)},e.mse=function(e,t){return w.meanSquaredError(e,t)},r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"meanAbsoluteError"})],e.prototype,"meanAbsoluteError",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"meanAbsolutePercentageError"})],e.prototype,"meanAbsolutePercentageError",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"binaryAccuracy"})],e,"binaryAccuracy",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"binaryCrossentropy"})],e,"binaryCrossentropy",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"categoricalAccuracy"})],e,"categoricalAccuracy",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"categoricalCrossentropy"})],e,"categoricalCrossentropy",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"cosineProximity"})],e,"cosineProximity",null),r([i.doc({heading:"Metrics",namespace:"metrics",useDocsFrom:"meanSquaredError"})],e,"meanSquaredError",null),e}();n.MetricExports=O;var k=function(){function e(){}return e.l1l2=function(e){return new E.L1L2(e)},e.l1=function(e){return E.l1(e)},e.l2=function(e){return E.l2(e)},r([i.doc({heading:"Regularizers",namespace:"regularizers",useDocsFrom:"L1L2"})],e,"l1l2",null),r([i.doc({heading:"Regularizers",namespace:"regularizers",useDocsFrom:"L1L2"})],e,"l1",null),r([i.doc({heading:"Regularizers",namespace:"regularizers",useDocsFrom:"L1L2"})],e,"l2",null),e}();n.RegularizerExports=k},{"./constraints":103,"./engine/topology":105,"./engine/training":106,"./initializers":110,"./layers/advanced_activations":111,"./layers/convolutional":112,"./layers/convolutional_depthwise":113,"./layers/core":114,"./layers/embeddings":115,"./layers/merge":116,"./layers/normalization":117,"./layers/padding":118,"./layers/pooling":119,"./layers/recurrent":120,"./layers/wrappers":122,"./losses":123,"./metrics":124,"./models":125,"./regularizers":127,"@tensorflow/tfjs-core":8}],109:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("./backend/tfjs_backend");n.backend=r;var i=e("./exports"),a=e("./callbacks");n.Callback=a.Callback,n.CallbackList=a.CallbackList,n.CustomCallback=a.CustomCallback;var o=e("./engine/training");n.Model=o.Model;var s=e("./layers/recurrent");n.RNN=s.RNN;var u=e("./models");n.Sequential=u.Sequential;var l=e("./types");n.SymbolicTensor=l.SymbolicTensor;var c=e("./version");n.version_layers=c.version,n.model=i.ModelExports.model,n.sequential=i.ModelExports.sequential,n.loadModel=i.ModelExports.loadModel,n.input=i.ModelExports.input,n.layers=i.LayerExports,n.constraints=i.ConstraintExports,n.initializers=i.InitializerExports,n.metrics=i.MetricExports,n.regularizers=i.RegularizerExports},{"./backend/tfjs_backend":100,"./callbacks":101,"./engine/training":106,"./exports":108,"./layers/recurrent":120,"./models":125,"./types":128,"./version":133}],110:[function(e,t,n){"use strict";function r(e){if(null!=e&&n.VALID_FAN_MODE_VALUES.indexOf(e)<0)throw new h.ValueError(e+" is not a valid FanMode.  Valid values as "+n.VALID_FAN_MODE_VALUES)}function i(e){if(null!=e&&n.VALID_DISTRIBUTION_VALUES.indexOf(e)<0)throw new h.ValueError(e+" is not a valid Distribution.  Valid values as "+n.VALID_DISTRIBUTION_VALUES)}function a(e,t){void 0===t&&(t="channelsLast");var n,r;if(p.checkDataFormat(t),2===e.length)n=e[0],r=e[1];else if(-1!==[3,4,5].indexOf(e.length)){if("channelsFirst"===t){i=g.arrayProd(e,2);n=e[1]*i,r=e[0]*i}else if("channelsLast"===t){var i=g.arrayProd(e,0,e.length-2);n=e[e.length-2]*i,r=e[e.length-1]*i}}else{var a=g.arrayProd(e);n=Math.sqrt(a),r=Math.sqrt(a)}return[n,r]}function o(e,t){return void 0===t&&(t={}),f.deserializeKerasObject(e,f.ClassNameMap.getMap().pythonClassNameMap,t,"initializer")}var s=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),u=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var l=e("@tensorflow/tfjs-core"),c=e("./backend/tfjs_backend"),p=e("./common"),h=e("./errors"),d=e("./types"),f=e("./utils/generic_utils"),g=e("./utils/math_utils");f.SerializableEnumRegistry.register("mode",{fan_in:"fanIn",fan_out:"fanOut",fan_avg:"fanAvg"}),n.VALID_FAN_MODE_VALUES=["fanIn","fanOut","fanAvg",void 0,null],n.checkFanMode=r,f.SerializableEnumRegistry.register("distribution",{normal:"normal",uniform:"uniform"}),n.VALID_DISTRIBUTION_VALUES=["normal","uniform",void 0,null],n.checkDistribution=i;var m=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return s(t,e),t.fromConfig=function(e,t){return new e(t)},t.prototype.fromConfigUsesCustomObjects=function(){return!1},t.prototype.getConfig=function(){return{}},t=u([l.doc({heading:"Initializers",subheading:"Classes",namespace:"initializers"})],t)}(d.Serializable);n.Initializer=m;var v=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return s(t,e),t.prototype.getClassName=function(){return"Zeros"},t.prototype.apply=function(e,t){return c.zeros(e,t)},t}(m);n.Zeros=v,f.ClassNameMap.register("Zeros",v);var y=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return s(t,e),t.prototype.getClassName=function(){return"Ones"},t.prototype.apply=function(e,t){return c.ones(e,t)},t}(m);n.Ones=y,f.ClassNameMap.register("Ones",y);var b=function(e){function t(t){var n=e.call(this)||this;return n.value=t.value,n}return s(t,e),t.prototype.apply=function(e,t){return c.scalarTimesArray(l.scalar(this.value),c.ones(e,t))},t.prototype.getClassName=function(){return"Constant"},t.prototype.getConfig=function(){return{value:this.value}},t}(m);n.Constant=b,f.ClassNameMap.register("Constant",b);var w=function(e){function t(t){var n=e.call(this)||this;return n.DEFAULT_MINVAL=-.05,n.DEFAULT_MAXVAL=.05,n.minval=t.minval||n.DEFAULT_MINVAL,n.maxval=t.maxval||n.DEFAULT_MAXVAL,n.seed=t.seed,n}return s(t,e),t.prototype.apply=function(e,t){return c.randomUniform(e,this.minval,this.maxval,t,this.seed)},t.prototype.getClassName=function(){return"RandomUniform"},t.prototype.getConfig=function(){return{minval:this.minval,maxval:this.maxval,seed:this.seed}},t}(m);n.RandomUniform=w,f.ClassNameMap.register("RandomUniform",w);var x=function(e){function t(t){var n=e.call(this)||this;return n.DEFAULT_MEAN=0,n.DEFAULT_STDDEV=.05,n.mean=t.mean||n.DEFAULT_MEAN,n.stddev=t.stddev||n.DEFAULT_STDDEV,n.seed=t.seed,n}return s(t,e),t.prototype.apply=function(e,t){return c.randomNormal(e,this.mean,this.stddev,t,this.seed)},t.prototype.getClassName=function(){return"RandomNormal"},t.prototype.getConfig=function(){return{mean:this.mean,stddev:this.stddev,seed:this.seed}},t}(m);n.RandomNormal=x,f.ClassNameMap.register("RandomNormal",x);var A=function(e){function t(t){var n=e.call(this)||this;return n.DEFAULT_MEAN=0,n.DEFAULT_STDDEV=.05,n.mean=t.mean||n.DEFAULT_MEAN,n.stddev=t.stddev||n.DEFAULT_STDDEV,n.seed=t.seed,n}return s(t,e),t.prototype.apply=function(e,t){return c.truncatedNormal(e,this.mean,this.stddev,t,this.seed)},t.prototype.getClassName=function(){return"TruncatedNormal"},t.prototype.getConfig=function(){return{mean:this.mean,stddev:this.stddev,seed:this.seed}},t}(m);n.TruncatedNormal=A,f.ClassNameMap.register("TruncatedNormal",A);var E=function(e){function t(t){var n=e.call(this)||this;return n.gain=null!=t.gain?l.scalar(t.gain):c.getScalar(1),n}return s(t,e),t.prototype.apply=function(e,t){if(2!==e.length||e[0]!==e[1])throw new h.ValueError("Identity matrix initializer can only be used for 2D square matrices.");return c.scalarTimesArray(this.gain,c.eye(e[0]))},t.prototype.getClassName=function(){return"Identity"},t.prototype.getConfig=function(){return{gain:this.gain.get()}},t}(m);n.Identity=E,f.ClassNameMap.register("Identity",E);var _=function(e){function t(t){var n=e.call(this)||this;if(t.scale<0)throw new h.ValueError("scale must be a positive float. Got: "+t.scale);return n.scale=null==t.scale?1:t.scale,n.mode=t.mode,r(n.mode),n.distribution=t.distribution,i(n.distribution),n.seed=t.seed,n}return s(t,e),t.prototype.apply=function(e,t){var n=a(e),r=n[0],i=n[1],o=this.scale;if("fanIn"===this.mode?o/=Math.max(1,r):"fanOut"===this.mode?o/=Math.max(1,i):o/=Math.max(1,(r+i)/2),"normal"===this.distribution){var s=Math.sqrt(o);return c.truncatedNormal(e,0,s,t,this.seed)}var u=Math.sqrt(3*o);return c.randomUniform(e,-u,u,t,this.seed)},t.prototype.getClassName=function(){return"VarianceScaling"},t.prototype.getConfig=function(){return{scale:this.scale,mode:this.mode,distribution:this.distribution,seed:this.seed}},t}(m);n.VarianceScaling=_,f.ClassNameMap.register("VarianceScaling",_);var T=function(e){function t(t){return e.call(this,{scale:1,mode:"fanAvg",distribution:"uniform",seed:t.seed})||this}return s(t,e),t}(_);n.GlorotUniform=T,f.ClassNameMap.register("GlorotUniform",T);var S=function(e){function t(t){return e.call(this,{scale:1,mode:"fanAvg",distribution:"normal",seed:t.seed})||this}return s(t,e),t}(_);n.GlorotNormal=S,f.ClassNameMap.register("GlorotNormal",S);var C=function(e){function t(t){return e.call(this,{scale:2,mode:"fanIn",distribution:"normal",seed:t.seed})||this}return s(t,e),t}(_);n.HeNormal=C,f.ClassNameMap.register("HeNormal",C);var O=function(e){function t(t){return e.call(this,{scale:1,mode:"fanIn",distribution:"normal",seed:t.seed})||this}return s(t,e),t}(_);n.LeCunNormal=O,f.ClassNameMap.register("LeCunNormal",O);var k=function(e){function t(t){var n=e.call(this)||this;if(n.DEFAULT_GAIN=1,n.gain=null==t.gain?n.DEFAULT_GAIN:t.gain,n.seed=t.seed,null!=n.seed)throw new h.NotImplementedError("Random seed is not implemented for Orthogonal Initializer yet.");return n}return s(t,e),t.prototype.apply=function(e,t){if(2!==e.length)throw new h.NotImplementedError("The Orthogonal Initializer does not support non-2D shapes yet.");var n=e[0]>=e[1]?e:[e[1],e[0]],r=c.randomNormal(n,0,1,d.DType.float32),i=c.qr(r)[0];return i.shape[1]>n[1]&&(i=i.slice([0,0],n)),e[0]<e[1]&&(i=i.transpose()),c.scalarTimesArray(c.getScalar(this.gain),i)},t.prototype.getClassName=function(){return"Orthogonal"},t.prototype.getConfig=function(){return{gain:this.gain,seed:this.seed}},t}(m);n.Orthogonal=k,f.ClassNameMap.register("Orthogonal",k),n.INITIALIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP={constant:"Constant",glorotNormal:"GlorotNormal",glorotUniform:"GlorotUniform",heNormal:"HeNormal",identity:"Identity",leCunNormal:"LeCunNormal",ones:"Ones",orthogonal:"Orthogonal",randomNormal:"RandomNormal",randomUniform:"RandomUniform",truncatedNormal:"TruncatedNormal",varianceScaling:"VarianceScaling",zeros:"Zeros"},n.serializeInitializer=function(e){return f.serializeKerasObject(e)},n.getInitializer=function(e){return"string"==typeof e?o({className:e in n.INITIALIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP?n.INITIALIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP[e]:e,config:{}}):e instanceof m?e:o(e)}},{"./backend/tfjs_backend":100,"./common":102,"./errors":107,"./types":128,"./utils/generic_utils":130,"./utils/math_utils":131,"@tensorflow/tfjs-core":8}],111:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("../activations"),o=e("../backend/tfjs_backend"),s=e("../backend/tfjs_backend"),u=e("../engine/topology"),l=e("../errors"),c=e("../types"),p=e("../utils/generic_utils"),h=function(e){function t(t){var n=e.call(this,null==t?{}:t)||this;return n.DEFAULT_ALPHA=.3,null==t&&(t={}),n.alpha=null==t.alpha?n.DEFAULT_ALPHA:t.alpha,n}return r(t,e),t.prototype.call=function(e,t){var n=p.getExactlyOneTensor(e);return i.leakyRelu(n,this.alpha)},t.prototype.computeOutputShape=function(e){return e},t.prototype.getClassName=function(){return"LeakyReLU"},t.prototype.getConfig=function(){var t={alpha:this.alpha},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.LeakyReLU=h,p.ClassNameMap.register("LeakyReLU",h);var d=function(e){function t(t){var n=e.call(this,null==t?{}:t)||this;if(n.DEFAULT_ALPHA=1,null==t&&(t={}),null!=t.alpha&&t.alpha!==n.DEFAULT_ALPHA)throw new l.NotImplementedError("Non-default alpha value ("+t.alpha+") is not supported by the ELU layer yet.");return n.alpha=null==t.alpha?n.DEFAULT_ALPHA:t.alpha,n}return r(t,e),t.prototype.call=function(e,t){var n=p.getExactlyOneTensor(e);return i.elu(n)},t.prototype.computeOutputShape=function(e){return e},t.prototype.getClassName=function(){return"ELU"},t.prototype.getConfig=function(){var t={alpha:this.alpha},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.ELU=d,p.ClassNameMap.register("ELU",d);var f=function(e){function t(t){var n=e.call(this,null==t?{}:t)||this;return n.DEFAULT_THETA=1,null==t&&(t={}),n.theta=null==t.theta?n.DEFAULT_THETA:t.theta,n.thetaTensor=s.getScalar(n.theta),n}return r(t,e),t.prototype.call=function(e,t){var n=p.getExactlyOneTensor(e);return n.mul(o.cast(n.greater(this.thetaTensor),c.DType.float32))},t.prototype.computeOutputShape=function(e){return e},t.prototype.getClassName=function(){return"ThresholdedReLU"},t.prototype.getConfig=function(){var t={theta:this.theta},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.ThresholdedReLU=f,p.ClassNameMap.register("ThresholdedReLU",f);var g=function(e){function t(t){var n=e.call(this,null==t?{}:t)||this;return n.DEFAULT_AXIS=1,null==t&&(t={}),n.axis=null==t.theta?n.DEFAULT_AXIS:t.theta,n}return r(t,e),t.prototype.call=function(e,t){var n=p.getExactlyOneTensor(e);return a.softmax(n,this.axis)},t.prototype.computeOutputShape=function(e){return e},t.prototype.getClassName=function(){return"Softmax"},t.prototype.getConfig=function(){var t={axis:this.axis},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.Softmax=g,p.ClassNameMap.register("Softmax",g)},{"../activations":98,"../backend/tfjs_backend":100,"../engine/topology":105,"../errors":107,"../types":128,"../utils/generic_utils":130,"@tensorflow/tfjs-core":8}],112:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("../activations"),o=e("../backend/tfjs_backend"),s=e("../common"),u=e("../constraints"),l=e("../engine/topology"),c=e("../errors"),p=e("../initializers"),h=e("../regularizers"),d=e("../types"),f=e("../utils/conv_utils"),g=e("../utils/generic_utils"),m=function(e){function t(t,n){var r=e.call(this,n)||this;if(r.kernel=null,r.bias=null,r.DEFAULT_KERNEL_INITIALIZER="glorotNormal",r.DEFAULT_BIAS_INITIALIZER="zeros",r.rank=t,1!==r.rank&&2!==r.rank)throw new c.NotImplementedError("Convolution layer for rank other than 1 or 2 ("+r.rank+") is not implemented yet.");if(r.filters=n.filters,r.kernelSize=f.normalizeArray(n.kernelSize,t,"kernelSize"),r.strides=f.normalizeArray(null==n.strides?1:n.strides,t,"strides"),r.padding=null==n.padding?"valid":n.padding,s.checkPaddingMode(r.padding),r.dataFormat=null==n.dataFormat?"channelsLast":n.dataFormat,s.checkDataFormat(r.dataFormat),r.dilationRate=null==n.dilationRate?1:n.dilationRate,1===r.rank&&Array.isArray(r.dilationRate)&&1!==r.dilationRate.length)throw new c.ValueError("dilationRate must be a number or an array of a single number for 1D convolution, but received "+JSON.stringify(r.dilationRate));if(2===r.rank)if("number"==typeof r.dilationRate)r.dilationRate=[r.dilationRate,r.dilationRate];else if(2!==r.dilationRate.length)throw new c.ValueError("dilationRate must be a number or array of two numbers for 2D convolution, but received "+JSON.stringify(r.dilationRate));return r.activation=a.getActivation(n.activation),r.useBias=null==n.useBias||n.useBias,r.kernelInitializer=p.getInitializer(n.kernelInitializer||r.DEFAULT_KERNEL_INITIALIZER),r.biasInitializer=p.getInitializer(n.biasInitializer||r.DEFAULT_BIAS_INITIALIZER),r.kernelConstraint=u.getConstraint(n.kernelConstraint),r.biasConstraint=u.getConstraint(n.biasConstraint),r.kernelRegularizer=h.getRegularizer(n.kernelRegularizer),r.biasRegularizer=h.getRegularizer(n.biasRegularizer),r.activityRegularizer=h.getRegularizer(n.activityRegularizer),r}return r(t,e),t.prototype.build=function(e){e=g.getExactlyOneShape(e);var t="channelsFirst"===this.dataFormat?1:e.length-1;if(null==e[t])throw new c.ValueError("The channel dimension of the input should be defined. Found "+e[t]);var n=e[t],r=this.kernelSize.concat([n,this.filters]);this.kernel=this.addWeight("kernel",r,null,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.useBias&&(this.bias=this.addWeight("bias",[this.filters],null,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint)),this.inputSpec=[{ndim:this.rank+2,axes:(i={},i[t]=n,i)}],this.built=!0;var i},t.prototype.call=function(e,t){e=g.getExactlyOneTensor(e);var n,r=null==this.bias?null:this.bias.read();if(1===this.rank)n=o.conv1dWithBias(e,this.kernel.read(),r,this.strides[0],this.padding,this.dataFormat,this.dilationRate);else if(2===this.rank)n=o.conv2dWithBias(e,this.kernel.read(),r,this.strides,this.padding,this.dataFormat,this.dilationRate);else if(3===this.rank)throw new c.NotImplementedError("3D convolution is not implemented yet.");return null!=this.activation&&(n=this.activation(n)),n},t.prototype.computeOutputShape=function(e){e=g.getExactlyOneShape(e);for(var t=[],n="channelsLast"===this.dataFormat?e.slice(1,e.length-1):e.slice(2),r=0;r<n.length;++r){var i=f.convOutputLength(n[r],this.kernelSize[r],this.padding,this.strides[r],"number"==typeof this.dilationRate?this.dilationRate:this.dilationRate[r]);t.push(i)}var a=[e[0]];return"channelsLast"===this.dataFormat?(a=a.concat(t)).push(this.filters):(a.push(this.filters),a=a.concat(t)),a},t.prototype.getConfig=function(){var t={rank:this.rank,filters:this.filters,kernelSize:this.kernelSize,strides:this.strides,padding:this.padding,dataFormat:this.dataFormat,dilationRate:this.dilationRate,activation:a.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:p.serializeInitializer(this.kernelInitializer),biasInitializer:p.serializeInitializer(this.biasInitializer),kernelRegularizer:h.serializeRegularizer(this.kernelRegularizer),biasRegularizer:h.serializeRegularizer(this.biasRegularizer),activityRegularizer:h.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint)},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(l.Layer);n.Conv=m;var v=function(e){function t(t){return e.call(this,2,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Conv2D"},t.prototype.getConfig=function(){var t=e.prototype.getConfig.call(this);return delete t.rank,t},t}(m);n.Conv2D=v,g.ClassNameMap.register("Conv2D",v);var y=function(e){function t(t){var n=e.call(this,t)||this;if(n.inputSpec=[new l.InputSpec({ndim:4})],"same"!==n.padding&&"valid"!==n.padding)throw new c.ValueError("Conv2DTranspose currently supports only padding modes 'same' and 'valid', but received padding mode "+n.padding);return n}return r(t,e),t.prototype.getClassName=function(){return"Conv2DTranspose"},t.prototype.build=function(e){if(4!==(e=g.getExactlyOneShape(e)).length)throw new c.ValueError("Input should have rank 4; Received input shape: "+JSON.stringify(e));var t="channelsFirst"===this.dataFormat?1:e.length-1;if(null==e[t])throw new c.ValueError("The channel dimension of the inputs should be defined. Found `None`.");var n=e[t],r=this.kernelSize.concat([this.filters,n]);this.kernel=this.addWeight("kernel",r,d.DType.float32,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.useBias&&(this.bias=this.addWeight("bias",[this.filters],d.DType.float32,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint)),this.inputSpec=[new l.InputSpec({ndim:4,axes:(i={},i[t]=n,i)})],this.built=!0;var i},t.prototype.call=function(e,t){var n=this;return i.tidy(function(){var t=g.getExactlyOneTensor(e);if(4!==t.shape.length)throw new c.ValueError("Conv2DTranspose.call() expects input tensor to be rank-4, but received a tensor of rank-"+t.shape.length);var r,a,s=t.shape,u=s[0];"channelsFirst"===n.dataFormat?(r=2,a=3):(r=1,a=2);var l=s[r],p=s[a],h=n.kernelSize[0],d=n.kernelSize[1],m=n.strides[0],v=n.strides[1],y=[u,f.deconvLength(l,m,h,n.padding),f.deconvLength(p,v,d,n.padding),n.filters];"channelsLast"!==n.dataFormat&&(t=o.transpose(t,[0,2,3,1]));var b=i.conv2dTranspose(t,n.kernel.read(),y,n.strides,n.padding);return"channelsLast"!==n.dataFormat&&(b=o.transpose(b,[0,3,1,2])),null!=n.bias&&(b=o.biasAdd(b,n.bias.read(),n.dataFormat)),null!=n.activation&&(b=n.activation(b)),b})},t.prototype.computeOutputShape=function(e){var t,n,r,i=(e=g.getExactlyOneShape(e)).slice();"channelsFirst"===this.dataFormat?(t=1,n=2,r=3):(t=3,n=1,r=2);var a=this.kernelSize[0],o=this.kernelSize[1],s=this.strides[0],u=this.strides[1];return i[t]=this.filters,i[n]=f.deconvLength(i[n],s,a,this.padding),i[r]=f.deconvLength(i[r],u,o,this.padding),i},t.prototype.getConfig=function(){var t=e.prototype.getConfig.call(this);return delete t.dilationRate,t},t}(v);n.Conv2DTranspose=y,g.ClassNameMap.register("Conv2DTranspose",y);var b=function(e){function t(t,n){var r=e.call(this,t,n)||this;if(r.DEFAULT_DEPTHWISE_INITIALIZER="glorotUniform",r.DEFAULT_POINTWISE_INITIALIZER="glorotUniform",r.depthwiseKernel=null,r.pointwiseKernel=null,null==n.filters)throw new c.ValueError("The `filters` configuration field is required by SeparableConv, but is unspecified.");if(null!=n.kernelInitializer||null!=n.kernelRegularizer||null!=n.kernelConstraint)throw new c.ValueError("Fields kernelInitializer, kernelRegularizer and kernelConstraint are invalid for SeparableConv2D. Use depthwiseInitializer, depthwiseRegularizer, depthwiseConstraint, pointwiseInitializer, pointwiseRegularizer and pointwiseConstraint instead.");if(null!=n.padding&&"same"!==n.padding&&"valid"!==n.padding)throw new c.ValueError("SeparableConv"+r.rank+"D supports only padding modes: 'same' and 'valid', but received "+JSON.stringify(n.padding));return r.depthMultiplier=null==n.depthMultiplier?1:n.depthMultiplier,r.depthwiseInitializer=p.getInitializer(n.depthwiseInitializer||r.DEFAULT_DEPTHWISE_INITIALIZER),r.depthwiseRegularizer=h.getRegularizer(n.depthwiseRegularizer),r.depthwiseConstraint=u.getConstraint(n.depthwiseConstraint),r.pointwiseInitializer=p.getInitializer(n.depthwiseInitializer||r.DEFAULT_POINTWISE_INITIALIZER),r.pointwiseRegularizer=h.getRegularizer(n.pointwiseRegularizer),r.pointwiseConstraint=u.getConstraint(n.pointwiseConstraint),r}return r(t,e),t.prototype.build=function(e){if((e=g.getExactlyOneShape(e)).length<this.rank+2)throw new c.ValueError("Inputs to SeparableConv"+this.rank+"D should have rank "+(this.rank+2)+", but received input shape: "+JSON.stringify(e));var t="channelsFirst"===this.dataFormat?1:e.length-1;if(null==e[t]||e[t]<0)throw new c.ValueError("The channel dimension of the inputs should be defined, but found "+JSON.stringify(e[t]));for(var n=e[t],r=this.kernelSize.concat([n,this.depthMultiplier]),i=[],a=0;a<this.rank;++a)i.push(1);i.push(n*this.depthMultiplier,this.filters);this.depthwiseKernel=this.addWeight("depthwise_kernel",r,d.DType.float32,this.depthwiseInitializer,this.depthwiseRegularizer,!0,this.depthwiseConstraint),this.pointwiseKernel=this.addWeight("pointwise_kernel",i,d.DType.float32,this.pointwiseInitializer,this.pointwiseRegularizer,!0,this.pointwiseConstraint),this.useBias?this.bias=this.addWeight("bias",[this.filters],d.DType.float32,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint):this.bias=null,this.inputSpec=[new l.InputSpec({ndim:this.rank+2,axes:(o={},o[t]=n,o)})],this.built=!0;var o},t.prototype.call=function(e,t){e=g.getExactlyOneTensor(e);var n;if(1===this.rank)throw new c.NotImplementedError("1D separable convolution is not implemented yet.");return 2===this.rank&&("channelsFirst"===this.dataFormat&&(e=o.transpose(e,[0,2,3,1])),n=i.separableConv2d(e,this.depthwiseKernel.read(),this.pointwiseKernel.read(),this.strides,this.padding,this.dilationRate,"NHWC")),this.useBias&&(n=o.biasAdd(n,this.bias.read(),this.dataFormat)),null!=this.activation&&(n=this.activation(n)),"channelsFirst"===this.dataFormat&&(n=o.transpose(n,[0,3,1,2])),n},t.prototype.getClassName=function(){return"SeparableConv"},t.prototype.getConfig=function(){var t=e.prototype.getConfig.call(this);return delete t.rank,delete t.kernelInitializer,delete t.kernelRegularizer,delete t.kernelConstraint,t.depthwiseInitializer=p.serializeInitializer(this.depthwiseInitializer),t.pointwiseInitializer=p.serializeInitializer(this.pointwiseInitializer),t.depthwiseRegularizer=h.serializeRegularizer(this.depthwiseRegularizer),t.pointwiseRegularizer=h.serializeRegularizer(this.pointwiseRegularizer),t.depthwiseConstraint=u.serializeConstraint(this.depthwiseConstraint),t.pointwiseConstraint=u.serializeConstraint(this.pointwiseConstraint),t},t}(m);n.SeparableConv=b;var w=function(e){function t(t){return e.call(this,2,t)||this}return r(t,e),t.prototype.getClassName=function(){return"SeparableConv2D"},t}(b);n.SeparableConv2D=w,g.ClassNameMap.register("SeparableConv2D",w);var x=function(e){function t(t){var n=e.call(this,1,t)||this;return n.inputSpec=[{ndim:3}],n}return r(t,e),t.prototype.getClassName=function(){return"Conv1D"},t.prototype.getConfig=function(){var t=e.prototype.getConfig.call(this);return delete t.rank,delete t.dataFormat,t},t}(m);n.Conv1D=x,g.ClassNameMap.register("Conv1D",x)},{"../activations":98,"../backend/tfjs_backend":100,"../common":102,"../constraints":103,"../engine/topology":105,"../errors":107,"../initializers":110,"../regularizers":127,"../types":128,"../utils/conv_utils":129,"../utils/generic_utils":130,"@tensorflow/tfjs-core":8}],113:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../backend/tfjs_backend"),a=e("../constraints"),o=e("../errors"),s=e("../initializers"),u=e("../regularizers"),l=e("../utils/conv_utils"),c=e("../utils/generic_utils"),p=e("../utils/generic_utils"),h=function(e){function t(t){var n=e.call(this,t)||this;return n.depthwiseKernel=null,n.depthMultiplier=null==t.depthMultiplier?1:t.depthMultiplier,n.depthwiseInitializer=s.getInitializer(t.depthwiseInitializer||n.DEFAULT_KERNEL_INITIALIZER),n.depthwiseConstraint=a.getConstraint(t.depthwiseConstraint),n.depthwiseRegularizer=u.getRegularizer(t.depthwiseRegularizer),n}return r(t,e),t.prototype.getClassName=function(){return"DepthwiseConv2D"},t.prototype.build=function(e){if((e=p.getExactlyOneShape(e)).length<4)throw new o.ValueError("Inputs to DepthwiseConv2D should have rank 4. Received input shape: "+JSON.stringify(e)+".");var t="channelsFirst"===this.dataFormat?1:3;if(null==e[t]||e[t]<0)throw new o.ValueError("The channel dimension of the inputs to DepthwiseConv2D should be defined, but is not ("+e[t]+").");var n=e[t],r=[this.kernelSize[0],this.kernelSize[1],n,this.depthMultiplier];this.depthwiseKernel=this.addWeight("depthwise_kernel",r,null,this.depthwiseInitializer,this.depthwiseRegularizer,!0,this.depthwiseConstraint),this.useBias?this.bias=this.addWeight("bias",[n*this.depthMultiplier],null,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint):this.bias=null,this.built=!0},t.prototype.call=function(e,t){e=p.getExactlyOneTensor(e);var n=i.depthwiseConv2d(e,this.depthwiseKernel.read(),this.strides,this.padding,this.dataFormat,null);return this.useBias&&(n=i.biasAdd(n,this.bias.read(),this.dataFormat)),null!=this.activation&&(n=this.activation(n)),n},t.prototype.computeOutputShape=function(e){e=p.getExactlyOneShape(e);var t="channelsFirst"===this.dataFormat?e[2]:e[1],n="channelsFirst"===this.dataFormat?e[3]:e[2],r="channelsFirst"===this.dataFormat?e[1]*this.depthMultiplier:e[3]*this.depthMultiplier,i=l.convOutputLength(t,this.kernelSize[0],this.padding,this.strides[0]),a=l.convOutputLength(n,this.kernelSize[1],this.padding,this.strides[1]);return"channelsFirst"===this.dataFormat?[e[0],r,i,a]:[e[0],i,a,r]},t}(e("./convolutional").Conv2D);n.DepthwiseConv2D=h,c.ClassNameMap.register("DepthwiseConv2D",h)},{"../backend/tfjs_backend":100,"../constraints":103,"../errors":107,"../initializers":110,"../regularizers":127,"../utils/conv_utils":129,"../utils/generic_utils":130,"./convolutional":112}],114:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("../activations"),o=e("../backend/tfjs_backend"),s=e("../constraints"),u=e("../engine/topology"),l=e("../errors"),c=e("../initializers"),p=e("../regularizers"),h=e("../utils/generic_utils"),d=e("../utils/generic_utils"),f=e("../utils/math_utils"),g=function(e){function t(t){var n=e.call(this,t)||this;if(n.rate=Math.max(Math.min(t.rate,1),0),n.rateScalar=o.getScalar(n.rate),n.noiseShape=t.noiseShape,n.seed=t.seed,null!=n.seed)throw new l.NotImplementedError("Non-default seed is not implemented in Dropout layer yet: "+n.seed);return n.supportsMasking=!0,n}return r(t,e),t.prototype.getNoiseShape=function(e){if(null==this.noiseShape)return this.noiseShape;for(var t=e.shape,n=[],r=0;r<this.noiseShape.length;++r)n.push(null==this.noiseShape[r]?t[r]:this.noiseShape[r]);return n},t.prototype.call=function(e,t){var n=this;this.invokeCallHook(e,t);var r=h.getExactlyOneTensor(e);if(null!=this.noiseShape&&!i.util.arraysEqual(r.shape,this.noiseShape))throw new l.NotImplementedError("Non-default noise shape is not implemented in Dropout layer yet: "+JSON.stringify(this.noiseShape));if(0<this.rate&&this.rate<1){var a=null!=t.training&&t.training,s=this.getNoiseShape(r);return o.inTrainPhase(function(){return o.dropout(r,n.rateScalar,s,n.seed)},function(){return r},a)}return e},t.prototype.getClassName=function(){return"Dropout"},t.prototype.getConfig=function(){var t={rate:this.rate,noiseShape:this.noiseShape,seed:this.seed},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.Dropout=g,h.ClassNameMap.register("Dropout",g);var m=function(e){function t(t){var n=e.call(this,t)||this;if(n.activation=null,n.useBias=!0,n.kernel=null,n.bias=null,n.DEFAULT_KERNEL_INITIALIZER="glorotNormal",n.DEFAULT_BIAS_INITIALIZER="zeros",null==t.batchInputShape&&null==t.inputShape&&null!=t.inputDim){var r=null;null!=t.batchSize&&(r=t.batchSize),n.batchInputShape=[r,t.inputDim]}return n.units=t.units,n.activation=a.getActivation(t.activation),null!=t.useBias&&(n.useBias=t.useBias),n.kernelInitializer=c.getInitializer(t.kernelInitializer||n.DEFAULT_KERNEL_INITIALIZER),n.biasInitializer=c.getInitializer(t.biasInitializer||n.DEFAULT_BIAS_INITIALIZER),n.kernelConstraint=s.getConstraint(t.kernelConstraint),n.biasConstraint=s.getConstraint(t.biasConstraint),n.kernelRegularizer=p.getRegularizer(t.kernelRegularizer),n.biasRegularizer=p.getRegularizer(t.biasRegularizer),n.activityRegularizer=p.getRegularizer(t.activityRegularizer),n.inputSpec=[{minNDim:2}],n}return r(t,e),t.prototype.build=function(e){var t=(e=h.getExactlyOneShape(e))[e.length-1];null==this.kernel&&(this.kernel=this.addWeight("kernel",[t,this.units],null,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.useBias&&(this.bias=this.addWeight("bias",[this.units],null,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint))),this.inputSpec=[{minNDim:2,axes:(n={},n[-1]=t,n)}],this.built=!0;var n},t.prototype.computeOutputShape=function(e){var t=(e=h.getExactlyOneShape(e)).slice();return t[t.length-1]=this.units,t},t.prototype.call=function(e,t){this.invokeCallHook(e,t);var n=h.getExactlyOneTensor(e),r=o.dot(n,this.kernel.read());return null!=this.bias&&(r=o.biasAdd(r,this.bias.read())),null!=this.activation&&(r=this.activation(r)),r},t.prototype.getClassName=function(){return"Dense"},t.prototype.getConfig=function(){var t={units:this.units,activation:a.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:c.serializeInitializer(this.kernelInitializer),biasInitializer:c.serializeInitializer(this.biasInitializer),kernelRegularizer:p.serializeRegularizer(this.kernelRegularizer),biasRegularizer:p.serializeRegularizer(this.biasRegularizer),activityRegularizer:p.serializeRegularizer(this.activityRegularizer),kernelConstraint:s.serializeConstraint(this.kernelConstraint),biasConstraint:s.serializeConstraint(this.biasConstraint)},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.Dense=m,h.ClassNameMap.register("Dense",m);var v=function(e){function t(t){var n=e.call(this,t||{})||this;return n.inputSpec=[{minNDim:3}],n}return r(t,e),t.prototype.computeOutputShape=function(e){for(var t=0,n=(e=h.getExactlyOneShape(e)).slice(1);t<n.length;t++)if(null==n[t])throw new l.ValueError('The shape of the input to "Flatten" is not fully defined (got '+e.slice(1)+'). Make sure to pass a complete "input_shape" or "batch_input_shape" argument to the first layer in your model.');return[e[0],f.arrayProd(e,1)]},t.prototype.getClassName=function(){return"Flatten"},t.prototype.call=function(e,t){return this.invokeCallHook(e,t),o.batchFlatten(h.getExactlyOneTensor(e))},t}(u.Layer);n.Flatten=v,h.ClassNameMap.register("Flatten",v);var y=function(e){function t(t){var n=e.call(this,t)||this;return n.supportsMasking=!0,n.activation=a.getActivation(t.activation),n}return r(t,e),t.prototype.getClassName=function(){return"Activation"},t.prototype.call=function(e,t){this.invokeCallHook(e,t);var n=h.getExactlyOneTensor(e);return this.activation(n)},t}(u.Layer);n.Activation=y,h.ClassNameMap.register("Activation",y);var b=function(e){function t(t){var n=e.call(this,t)||this;return n.n=t.n,n.inputSpec=[{ndim:2}],n}return r(t,e),t.prototype.computeOutputShape=function(e){return[e[0],this.n,e[1]]},t.prototype.call=function(e,t){return e=d.getExactlyOneTensor(e),o.repeat(e,this.n)},t.prototype.getClassName=function(){return"RepeatVector"},t.prototype.getConfig=function(){var t={n:this.n},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(u.Layer);n.RepeatVector=b,h.ClassNameMap.register("RepeatVector",b);var w=function(e){function t(t){var n=e.call(this,t)||this;n.targetShape=t.targetShape;for(var r=0;r<n.targetShape.length;++r)n.isUnknown(n.targetShape[r])&&(n.targetShape[r]=null);return n}return r(t,e),t.prototype.isUnknown=function(e){return e<0||null==e},t.prototype.fixUnknownDimension=function(e,t){for(var n="Total size of new array must be unchanged.",r=t.slice(),i=1,a=null,o=0;o<r.length;++o){var s=r[o];if(this.isUnknown(s)){if(null!==a)throw new l.ValueError("Can only specifiy one unknown dimension.");a=o}else i*=s}var u=f.arrayProd(e);if(null!==a){if(0===i||u%i!=0)throw new l.ValueError(n);r[a]=u/i}else if(u!==i)throw new l.ValueError(n);return r},t.prototype.computeOutputShape=function(e){for(var t=!1,n=0;n<e.length;++n)if(this.isUnknown(e[n])){t=!0;break}return t?e.slice(0,1).concat(this.targetShape):e.slice(0,1).concat(this.fixUnknownDimension(e.slice(1),this.targetShape))},t.prototype.getClassName=function(){return"Reshape"},t.prototype.call=function(e,t){this.invokeCallHook(e,t);var n=h.getExactlyOneTensor(e),r=o.shape(n),i=r.slice(0,1).concat(this.fixUnknownDimension(r.slice(1),this.targetShape));return o.reshape(n,i)},t}(u.Layer);n.Reshape=w,h.ClassNameMap.register("Reshape",w)},{"../activations":98,"../backend/tfjs_backend":100,"../constraints":103,"../engine/topology":105,"../errors":107,"../initializers":110,"../regularizers":127,"../utils/generic_utils":130,"../utils/math_utils":131,"@tensorflow/tfjs-core":8}],115:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../backend/tfjs_backend"),a=e("../constraints"),o=e("../engine/topology"),s=e("../errors"),u=e("../initializers"),l=e("../regularizers"),c=e("../utils/generic_utils"),p=e("../utils/generic_utils"),h=function(e){function t(t){var n=e.call(this,t)||this;if(n.embeddings=null,n.DEFAULT_EMBEDDINGS_INITIALIZER="randomUniform",null==t.batchInputShape&&null==t.inputShape){var r=null;null!=t.batchSize&&(r=t.batchSize),null==t.inputLength?n.batchInputShape=[r,null]:n.batchInputShape=[r].concat(c.toList(t.inputLength))}return n.inputDim=t.inputDim,n.outputDim=t.outputDim,n.embeddingsInitializer=u.getInitializer(t.embeddingsInitializer||n.DEFAULT_EMBEDDINGS_INITIALIZER),n.embeddingsRegularizer=l.getRegularizer(t.embeddingsRegularizer),n.activityRegularizer=l.getRegularizer(t.activityRegularizer),n.embeddingsConstraint=a.getConstraint(t.embeddingsConstraint),n.maskZero=t.maskZero,n.inputLength=t.inputLength,n}return r(t,e),t.prototype.build=function(e){this.embeddings=this.addWeight("embeddings",[this.inputDim,this.outputDim],this.dtype,this.embeddingsInitializer,this.embeddingsRegularizer,!0,this.embeddingsConstraint),this.built=!0},t.prototype.computeMask=function(e,t){throw new s.NotImplementedError("computeMask has not been implemented for Embedding yet")},t.prototype.computeOutputShape=function(e){if(e=c.getExactlyOneShape(e),null==this.inputLength)return e.concat([this.outputDim]);var t=c.toList(this.inputLength);if(t.length!==e.length-1)throw new s.ValueError('"inputLength" is '+this.inputLength+", but received input shape has shape "+e);for(var n=0,r=0;r<t.length;++r){var i=t[r],a=e[r+1];if(null!=i&&null!=a&&i!==a)throw new s.ValueError('"inputLength" is '+this.inputLength+", but received input shape has shape "+e);null==i&&(t[n]=a),n++}return[e[0]].concat(t,[this.outputDim])},t.prototype.call=function(e,t){this.invokeCallHook(e,t);var n=c.getExactlyOneTensor(e);"int32"!==i.dtype(n)&&(n=i.cast(n,"int32"));var r=i.gather(this.embeddings.read(),n.as1D());return i.reshape(r,p.getExactlyOneShape(this.computeOutputShape(n.shape)))},t.prototype.getClassName=function(){return"Embedding"},t.prototype.getConfig=function(){var t={inputDim:this.inputDim,outputDim:this.outputDim,embeddingsInitializer:u.serializeInitializer(this.embeddingsInitializer),embeddingsRegularizer:l.serializeRegularizer(this.embeddingsRegularizer),activityRegularizer:l.serializeRegularizer(this.activityRegularizer),embeddingsConstraint:a.serializeConstraint(this.embeddingsConstraint),maskZero:this.maskZero,inputLength:this.inputLength},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(o.Layer);n.Embedding=h,c.ClassNameMap.register("Embedding",h)},{"../backend/tfjs_backend":100,"../constraints":103,"../engine/topology":105,"../errors":107,"../initializers":110,"../regularizers":127,"../utils/generic_utils":130}],116:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("../backend/tfjs_backend"),o=e("../engine/topology"),s=e("../errors"),u=e("../utils/generic_utils"),l=e("../utils/math_utils"),c=function(e){function t(t){var n=e.call(this,t||{})||this;return n.supportsMasking=!0,n}return r(t,e),t.prototype.getClassName=function(){return"Merge"},t.prototype.mergeFunction=function(e){throw new s.NotImplementedError},t.prototype.computeElementwiseOpOutputShape=function(e,t){if(null==e||null==t)return null;if(e.length<t.length)return this.computeElementwiseOpOutputShape(t,e);if(0===t.length)return e;for(var n=e.slice(0,e.length-t.length),r=0;r<t.length;++r){var i=e[e.length-t.length+r],a=t[r];if(null==i||null==a||i<0||a<0)n.push(null);else if(1===i)n.push(a);else if(1===a)n.push(i);else{if(i!==a)throw new s.ValueError("Operands could not be broadcast together with shapes "+JSON.stringify(e)+" "+JSON.stringify(t));n.push(i)}}return n},t.prototype.build=function(e){if(Array.isArray(e)&&!Array.isArray(e[0])&&(e=[u.getExactlyOneShape(e)]),(e=e).length<2)throw new s.ValueError("A merge layer should be called on an Array of at least 2 inputs. Got "+e.length+" input(s).");for(var t=[],n=0,r=e;n<r.length;n++)null!=(o=r[n])&&null!==o[0]&&t.push(o[0]);if((t=u.unique(t)).length>1)throw new s.ValueError("Can not merge tensors with different batch sizes. Got tensors with shapes: "+JSON.stringify(e)+".");for(var i=null==e[0]?null:e[0].slice(1),a=1;a<e.length;++a){var o=null==e[a]?null:e[a].slice(1);i=this.computeElementwiseOpOutputShape(i,o)}var l=e.map(function(e){return e.length});-1===e.indexOf(null)&&1===u.unique(l).length?this.reshapeRequired=!1:this.reshapeRequired=!0},t.prototype.call=function(e,t){if(e=e,this.reshapeRequired){var n=[],r=e.map(function(e){return a.ndim(e)});if(-1===r.indexOf(null)){for(var i=l.max(r),o=0,s=e;o<s.length;o++){for(var u=s[o],c=a.ndim(u),p=0;p<i-c;++p)u=a.expandDims(u,1);n.push(u)}return this.mergeFunction(n)}for(var h=!1,d=0,f=e;d<f.length;d++){u=f[d];if(null==(c=a.ndim(u))){var g=a.shape(u),m=g[0],v=g.slice(1).concat([m]),y=a.reshape(u,[m].concat(l.arrayProd(g.slice(1))));y=a.permuteDimensions(y,[1,0]),y=a.reshape(y,v),n.push(y),h=!0}else if(c>1){A=l.range(1,c).concat([0]);n.push(a.permuteDimensions(u,A)),h=!0}else n.push(u)}var b=this.mergeFunction(n),w=a.ndim(b);if(h)if(null==w){var x=a.shape(b),v=[m=x[x.length-1]].concat(x.slice(0,x.length-1));b=a.reshape(a.permuteDimensions(a.reshape(b,[-1,m]),[1,0]),v)}else if(w>1){var A=[w-1].concat(l.range(0,w-1));b=a.permuteDimensions(b,A)}return b}return this.mergeFunction(e)},t.prototype.computeOutputShape=function(e){var t;t=null==(e=e)[0]?null:e[0].slice(1);for(var n=1;n<e.length;++n){var r=null==e[n]?null:e[n].slice(1);t=this.computeElementwiseOpOutputShape(t,r)}for(var i=[],a=0,o=e;a<o.length;a++)null!=(r=o[a])&&null!==r[0]&&i.push(r[0]);return i=u.unique(i),t=1===i.length?i.concat(t):[null].concat(t)},t}(o.Layer);n.Merge=c;var p=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Add"},t.prototype.mergeFunction=function(e){for(var t=a.zeros(e[0].shape),n=0,r=e;n<r.length;n++){var i=r[n];t=a.add(t,i)}return t},t}(c);n.Add=p,u.ClassNameMap.register("Add",p),n.add=function(e){return Array.isArray(e)?new p({}).apply(e):new p(e)};var h=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Multiply"},t.prototype.mergeFunction=function(e){for(var t=a.ones(e[0].shape),n=0,r=e;n<r.length;n++){var i=r[n];t=a.multiply(t,i)}return t},t}(c);n.Multiply=h,u.ClassNameMap.register("Multiply",h),n.multiply=function(e){return Array.isArray(e)?new h({}).apply(e):new h(e)};var d=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Average"},t.prototype.mergeFunction=function(e){for(var t=a.zeros(e[0].shape),n=0,r=e;n<r.length;n++){var i=r[n];t=a.add(t,i)}return a.scalarTimesArray(a.getScalar(1/e.length),t)},t}(c);n.Average=d,u.ClassNameMap.register("Average",d),n.average=function(e){return Array.isArray(e)?new d({}).apply(e):new d(e)};var f=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Maximum"},t.prototype.mergeFunction=function(e){for(var t=e[0],n=1;n<e.length;++n)t=a.maximum(t,e[n]);return t},t}(c);n.Maximum=f,u.ClassNameMap.register("Maximum",f),n.maximum=function(e){return Array.isArray(e)?new f({}).apply(e):new f(e)};var g=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"Minimum"},t.prototype.mergeFunction=function(e){for(var t=e[0],n=1;n<e.length;++n)t=a.minimum(t,e[n]);return t},t}(c);n.Minimum=g,u.ClassNameMap.register("Minimum",g),n.minimum=function(e){return Array.isArray(e)?new g({}).apply(e):new g(e)};var m=function(e){function t(t){var n=e.call(this,t)||this;return n.DEFAULT_AXIS=-1,null==t&&(t={}),n.axis=null==t.axis?n.DEFAULT_AXIS:t.axis,n.supportsMasking=!0,n.reshapeRequired=!1,n}return r(t,e),t.prototype.getClassName=function(){return"Concatenate"},t.prototype.build=function(e){if(!Array.isArray(e)||!Array.isArray(e[0])||1===e.length)throw new s.ValueError("A `Concatenate` layer should be called on a list of at least 2 inputs");for(var t=!0,n=0,r=e=e;n<r.length;n++)if(null!=(h=r[n])){t=!1;break}if(!t){for(var a=[],o=0;o<e.length;++o){var u=e[o].slice();u.splice(this.axis,1);for(var l=!1,c=0,p=a;c<p.length;c++){var h=p[c];if(i.util.arraysEqual(h,u)){l=!0;break}}l||a.push(u)}if(a.length>1)throw new s.ValueError("A `Concatenate` layer requires inputs with matching shapes except for the concat axis. Got input shapes: "+JSON.stringify(e))}},t.prototype.mergeFunction=function(e){return a.concatenate(e,this.axis)},t.prototype.computeOutputShape=function(e){if(!Array.isArray(e)||!Array.isArray(e[0]))throw new s.ValueError("A `Concatenate` layer should be called on a list of inputs.");for(var t=e,n=t[0].slice(),r=this.axis<0?n.length+this.axis:this.axis,i=0,a=t.slice(1);i<a.length;i++){var o=a[i];if(null==n[r]||null==o[r]){n[r]=null;break}n[r]+=o[r]}return n},t}(c);n.Concatenate=m,u.ClassNameMap.register("Concatenate",m),n.concatenate=function(e){return Array.isArray(e)?new m({}).apply(e):new m(e)}},{"../backend/tfjs_backend":100,"../engine/topology":105,"../errors":107,"../utils/generic_utils":130,"../utils/math_utils":131,"@tensorflow/tfjs-core":8}],117:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("@tensorflow/tfjs-core"),a=e("../backend/tfjs_backend"),o=e("../constraints"),s=e("../engine/topology"),u=e("../errors"),l=e("../initializers"),c=e("../regularizers"),p=e("../utils/generic_utils"),h=e("../utils/math_utils"),d=function(e){function t(t){var n=e.call(this,t)||this;return n.supportsMasking=!0,n.axis=null==t.axis?-1:t.axis,n.momentum=null==t.momentum?.99:t.momentum,n.epsilon=null==t.epsilon?.001:t.epsilon,n.center=null==t.center||t.center,n.scale=null==t.scale||t.scale,n.betaInitializer=l.getInitializer(t.betaInitializer||"zeros"),n.gammaInitializer=l.getInitializer(t.gammaInitializer||"ones"),n.movingMeanInitializer=l.getInitializer(t.movingMeanInitializer||"zeros"),n.movingVarianceInitializer=l.getInitializer(t.movingVarianceInitializer||"ones"),n.betaConstraint=o.getConstraint(t.betaConstraint),n.gammaConstraint=o.getConstraint(t.gammaConstraint),n.betaRegularizer=c.getRegularizer(t.betaRegularizer),n.gammaRegularizer=c.getRegularizer(t.gammaRegularizer),n.stepCount=0,n}return r(t,e),t.prototype.build=function(e){e=p.getExactlyOneShape(e);var t=this.axis>=0?this.axis:this.axis+e.length,n=e[t];if(null==n)throw new u.ValueError("Axis "+t+" of input tensor should have a defined dimension but the layer received an input with shape "+JSON.stringify(e)+".");this.inputSpec=[new s.InputSpec({ndim:e.length,axes:(i={},i[t]=n,i)})];var r=[n];this.scale&&(this.gamma=this.addWeight("gamma",r,null,this.gammaInitializer,this.gammaRegularizer,!0,this.gammaConstraint)),this.center&&(this.beta=this.addWeight("beta",r,null,this.betaInitializer,this.betaRegularizer,!0,this.betaConstraint)),this.movingMean=this.addWeight("moving_mean",r,null,this.movingMeanInitializer,null,!1),this.movingVariance=this.addWeight("moving_variance",r,null,this.movingVarianceInitializer,null,!1),this.built=!0;var i},t.prototype.call=function(e,t){var n=this;return i.tidy(function(){var r=null!=t.training&&t.training,o=p.getExactlyOneTensor(e),s=a.shape(o),u=s.length,l=h.range(0,u),c=n.axis>=0?n.axis:n.axis+u;l.splice(c,1);var d=p.pyListRepeat(1,u);d[c]=s[c];var f=l.slice();f.sort();var g=!i.util.arraysEqual(f,h.range(0,u).slice(0,u-1));if(!r)return function(){if(g){var e=a.reshape(n.movingMean.read(),d),t=a.reshape(n.movingVariance.read(),d),r=n.center?a.reshape(n.beta.read(),d):null,i=n.scale?a.reshape(n.gamma.read(),d):null;return a.batchNormalization(o,e,t,r,i,n.epsilon)}return a.batchNormalization(o,n.movingMean.read(),n.movingVariance.read(),null==n.beta?null:n.beta.read(),null==n.gamma?null:n.gamma.read(),n.epsilon)}();var m=a.normalizeBatchInTraining(o,n.gamma.read(),n.beta.read(),l,n.epsilon),v=m[0],y=m[1],b=m[2],w=h.arrayProd(l.map(function(e){return o.shape[e]})),x=b.mul(a.getScalar(w/(w-(1+n.epsilon))));return function(){n.stepCount++;var e=i.movingAverage(n.movingMean.read(),y,n.momentum,n.stepCount);n.movingMean.write(e);var t=i.movingAverage(n.movingVariance.read(),x,n.momentum,n.stepCount);n.movingVariance.write(t)}(),v})},t.prototype.getClassName=function(){return"BatchNormalization"},t.prototype.getConfig=function(){var t={axis:this.axis,momentum:this.momentum,epsilon:this.epsilon,center:this.center,scale:this.scale,betaInitializer:l.serializeInitializer(this.betaInitializer),gammaInitializer:l.serializeInitializer(this.gammaInitializer),movingMeanInitializer:l.serializeInitializer(this.movingMeanInitializer),movingVarianceInitializer:l.serializeInitializer(this.movingVarianceInitializer),betaRegularizer:c.serializeRegularizer(this.betaRegularizer),gammaRegularizer:c.serializeRegularizer(this.gammaRegularizer),betaConstraint:o.serializeConstraint(this.betaConstraint),gammaConstraint:o.serializeConstraint(this.gammaConstraint)},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(s.Layer);n.BatchNormalization=d,p.ClassNameMap.register("BatchNormalization",d)},{"../backend/tfjs_backend":100,"../constraints":103,"../engine/topology":105,"../errors":107,"../initializers":110,"../regularizers":127,"../utils/generic_utils":130,"../utils/math_utils":131,"@tensorflow/tfjs-core":8}],118:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../backend/common"),a=e("../backend/tfjs_backend"),o=e("../engine/topology"),s=e("../errors"),u=e("../utils/generic_utils"),l=function(e){function t(t){var n=this;if(null==t&&(t={}),n=e.call(this,t)||this,n.dataFormat=null==t.dataFormat?i.imageDataFormat():t.dataFormat,null==t.padding)n.padding=[[1,1],[1,1]];else if("number"==typeof t.padding)n.padding=[[t.padding,t.padding],[t.padding,t.padding]];else{if(t.padding=t.padding,2!==t.padding.length)throw new s.ValueError("ZeroPadding2D expects padding to be a length-2 array, but received a length-"+t.padding.length+" array.");var r=void 0,a=void 0;if("number"==typeof t.padding[0])r=[t.padding[0],t.padding[0]],a=[t.padding[1],t.padding[1]];else{if(t.padding=t.padding,2!==t.padding[0].length)throw new s.ValueError("ZeroPadding2D expects height padding to be a length-2 array, but received a length-"+t.padding[0].length+" array.");if(r=t.padding[0],2!==t.padding[1].length)throw new s.ValueError("ZeroPadding2D expects width padding to be a length-2 array, but received a length-"+t.padding[1].length+" array.");a=t.padding[1]}n.padding=[r,a]}return n.inputSpec=[new o.InputSpec({ndim:4})],n}return r(t,e),t.prototype.computeOutputShape=function(e){e=u.getExactlyOneShape(e);var t,n;return"channelsFirst"===this.dataFormat?(t=null!=e[2]&&e[2]>=0?e[2]+this.padding[0][0]+this.padding[0][1]:null,n=null!=e[3]&&e[3]>=0?e[3]+this.padding[1][0]+this.padding[1][1]:null,[e[0],e[1],t,n]):(t=null!=e[1]&&e[1]>=0?e[1]+this.padding[0][0]+this.padding[0][1]:null,n=null!=e[2]&&e[2]>=0?e[2]+this.padding[1][0]+this.padding[1][1]:null,[e[0],t,n,e[3]])},t.prototype.call=function(e,t){return a.spatial2dPadding(u.getExactlyOneTensor(e),this.padding,this.dataFormat)},t.prototype.getClassName=function(){return"ZeroPadding2D"},t.prototype.getConfig=function(){var t={padding:this.padding,dataFormat:this.dataFormat},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(o.Layer);n.ZeroPadding2D=l,u.ClassNameMap.register("ZeroPadding2D",l)},{"../backend/common":99,"../backend/tfjs_backend":100,"../engine/topology":105,"../errors":107,"../utils/generic_utils":130}],119:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../backend/tfjs_backend"),a=e("../common"),o=e("../engine/topology"),s=e("../engine/topology"),u=e("../errors"),l=e("../utils/conv_utils"),c=e("../utils/generic_utils"),p=function(e){function t(t){var n=this;return null==t.poolSize&&(t.poolSize=2),n=e.call(this,t)||this,n.poolSize=[t.poolSize],n.strides=null==t.strides?n.poolSize:[t.strides],n.padding=null==t.padding?"valid":t.padding,a.checkPaddingMode(n.padding),n.inputSpec=[new o.InputSpec({ndim:3})],n}return r(t,e),t.prototype.computeOutputShape=function(e){return e=c.getExactlyOneShape(e),length=l.convOutputLength(e[1],this.poolSize[0],this.padding,this.strides[0]),[e[0],length,e[2]]},t.prototype.call=function(e,t){this.invokeCallHook(e,t),e=i.expandDims(c.getExactlyOneTensor(e),2);var n=this.poolingFunction(c.getExactlyOneTensor(e),[this.poolSize[0],1],[this.strides[0],1],this.padding,"channelsLast");return i.squeeze(n,2)},t.prototype.getConfig=function(){var t={poolSize:this.poolSize,padding:this.padding,strides:this.strides},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(s.Layer);n.Pooling1D=p;var h=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"MaxPooling1D"},t.prototype.poolingFunction=function(e,t,n,r,o){return a.checkDataFormat(o),a.checkPaddingMode(r),i.pool2d(e,t,n,r,o,"max")},t}(p);n.MaxPooling1D=h,c.ClassNameMap.register("MaxPooling1D",h);var d=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"AveragePooling1D"},t.prototype.poolingFunction=function(e,t,n,r,o){return a.checkDataFormat(o),a.checkPaddingMode(r),i.pool2d(e,t,n,r,o,"avg")},t}(p);n.AveragePooling1D=d,c.ClassNameMap.register("AveragePooling1D",d);var f=function(e){function t(t){var n=this;return null==t.poolSize&&(t.poolSize=[2,2]),n=e.call(this,t)||this,n.poolSize=Array.isArray(t.poolSize)?t.poolSize:[t.poolSize,t.poolSize],n.strides=null==t.strides?n.poolSize:t.strides,n.padding=null==t.padding?"valid":t.padding,n.dataFormat=null==t.dataFormat?"channelsLast":t.dataFormat,a.checkDataFormat(n.dataFormat),a.checkPaddingMode(n.padding),n.inputSpec=[new o.InputSpec({ndim:4})],n}return r(t,e),t.prototype.computeOutputShape=function(e){e=c.getExactlyOneShape(e);var t="channelsFirst"===this.dataFormat?e[2]:e[1],n="channelsFirst"===this.dataFormat?e[3]:e[2];return t=l.convOutputLength(t,this.poolSize[0],this.padding,this.strides[0]),n=l.convOutputLength(n,this.poolSize[1],this.padding,this.strides[1]),"channelsFirst"===this.dataFormat?[e[0],e[1],t,n]:[e[0],t,n,e[3]]},t.prototype.call=function(e,t){return this.invokeCallHook(e,t),this.poolingFunction(c.getExactlyOneTensor(e),this.poolSize,this.strides,this.padding,this.dataFormat)},t.prototype.getConfig=function(){var t={poolSize:this.poolSize,padding:this.padding,strides:this.strides,dataFormat:this.dataFormat},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(s.Layer);n.Pooling2D=f;var g=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"MaxPooling2D"},t.prototype.poolingFunction=function(e,t,n,r,o){return a.checkDataFormat(o),a.checkPaddingMode(r),i.pool2d(e,t,n,r,o,"max")},t}(f);n.MaxPooling2D=g,c.ClassNameMap.register("MaxPooling2D",g);var m=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"AveragePooling2D"},t.prototype.poolingFunction=function(e,t,n,r,o){return a.checkDataFormat(o),a.checkPaddingMode(r),i.pool2d(e,t,n,r,o,"avg")},t}(f);n.AveragePooling2D=m,c.ClassNameMap.register("AveragePooling2D",m);var v=function(e){function t(t){var n=e.call(this,t)||this;return n.inputSpec=[new o.InputSpec({ndim:3})],n}return r(t,e),t.prototype.computeOutputShape=function(e){return[e[0],e[2]]},t.prototype.call=function(e,t){throw new u.NotImplementedError},t}(s.Layer);n.GlobalPooling1D=v;var y=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"GlobalAveragePooling1D"},t.prototype.call=function(e,t){var n=c.getExactlyOneTensor(e);return i.mean(n,1)},t}(v);n.GlobalAveragePooling1D=y,c.ClassNameMap.register("GlobalAveragePooling1D",y);var b=function(e){function t(t){return e.call(this,t)||this}return r(t,e),t.prototype.getClassName=function(){return"GlobalMaxPooling1D"},t.prototype.call=function(e,t){var n=c.getExactlyOneTensor(e);return i.max(n,1)},t}(v);n.GlobalMaxPooling1D=b,c.ClassNameMap.register("GlobalMaxPooling1D",b);var w=function(e){function t(t){var n=e.call(this,t)||this;return n.dataFormat=null==t.dataFormat?"channelsLast":t.dataFormat,a.checkDataFormat(n.dataFormat),n.inputSpec=[new o.InputSpec({ndim:4})],n}return r(t,e),t.prototype.computeOutputShape=function(e){return e=e,"channelsLast"===this.dataFormat?[e[0],e[3]]:[e[0],e[1]]},t.prototype.call=function(e,t){throw new u.NotImplementedError},t.prototype.getConfig=function(){var t={dataFormat:this.dataFormat},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(s.Layer);n.GlobalPooling2D=w;var x=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return r(t,e),t.prototype.call=function(e,t){var n=c.getExactlyOneTensor(e);return"channelsLast"===this.dataFormat?i.mean(n,[1,2]):i.mean(n,[2,3])},t.prototype.getClassName=function(){return"GlobalAveragePooling2D"},t}(w);n.GlobalAveragePooling2D=x,c.ClassNameMap.register("GlobalAveragePooling2D",x);var A=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return r(t,e),t.prototype.call=function(e,t){var n=c.getExactlyOneTensor(e);return"channelsLast"===this.dataFormat?i.max(n,[1,2]):i.max(n,[2,3])},t.prototype.getClassName=function(){return"GlobalMaxPooling2D"},t}(w);n.GlobalMaxPooling2D=A,c.ClassNameMap.register("GlobalMaxPooling2D",A)},{"../backend/tfjs_backend":100,"../common":102,"../engine/topology":105,"../errors":107,"../utils/conv_utils":129,"../utils/generic_utils":130}],120:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),i=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var a=e("@tensorflow/tfjs-core"),o=e("../activations"),s=e("../backend/tfjs_backend"),u=e("../constraints"),l=e("../engine/topology"),c=e("../engine/topology"),p=e("../errors"),h=e("../initializers"),d=e("../regularizers"),f=e("../types"),g=e("../utils/generic_utils"),m=e("../utils/math_utils"),v=e("./serialization"),y=function(e){function t(t){var n,r=e.call(this,t)||this;if(null==t.cell)throw new p.ValueError("cell property is missing for the constructor of RNN.");if(null==(n=Array.isArray(t.cell)?new S({cells:t.cell}):t.cell).stateSize)throw new p.ValueError("The RNN cell should have an attribute `stateSize` (tuple of integers, one integer per RNN state).");return r.cell=n,r.returnSequences=null!=t.returnSequences&&t.returnSequences,r.returnState=null!=t.returnState&&t.returnState,r.goBackwards=null!=t.goBackwards&&t.goBackwards,r._stateful=null!=t.stateful&&t.stateful,r.unroll=null!=t.unroll&&t.unroll,r.supportsMasking=!0,r.inputSpec=[new l.InputSpec({ndim:3})],r.stateSpec=null,r.states=null,r.numConstants=null,r}return r(t,e),t.prototype.getStates=function(){if(null==this.states){var e=Array.isArray(this.cell.stateSize)?this.cell.stateSize.length:1;return m.range(0,e).map(function(e){return null})}return this.states},t.prototype.setStates=function(e){this.states=e},t.prototype.computeOutputShape=function(e){g.isArrayOfShapes(e)&&(e=e[0]),e=e;var t=this.cell.stateSize;Array.isArray(t)||(t=[t]);var n,r=t[0];if(n=this.returnSequences?[e[0],e[1],r]:[e[0],r],this.returnState){for(var i=[],a=0,o=t;a<o.length;a++){var s=o[a];i.push([e[0],s])}return[n].concat(i)}return n},t.prototype.computeMask=function(e,t){throw new p.NotImplementedError("computeMask has not been implemented for RNN yet")},t.prototype.build=function(e){if(null!=this.numConstants)throw new p.NotImplementedError("Constants support is not implemented in RNN yet.");g.isArrayOfShapes(e)&&(e=e[0]),e=e;var t=this.stateful?e[0]:null,n=e[e.length-1];this.inputSpec[0]=new l.InputSpec({shape:[t,null,n]});var r=[e[0]].concat(e.slice(2));this.cell.build(r);var i;if(i=Array.isArray(this.cell.stateSize)?this.cell.stateSize:[this.cell.stateSize],null!=this.stateSpec){if(!a.util.arraysEqual(this.stateSpec.map(function(e){return e.shape[e.shape.length-1]}),i))throw new p.ValueError("An initialState was passed that is not compatible with cell.stateSize. Received stateSpec="+this.stateSpec+"; However cell.stateSize is "+this.cell.stateSize)}else this.stateSpec=i.map(function(e){return new l.InputSpec({shape:[null,e]})});if(this.stateful)throw new p.NotImplementedError("stateful RNN layer is not implemented yet")},t.prototype.resetStates=function(e){if(!this.stateful)throw new p.AttributeError("Cannot call resetState() on an RNN Layer that is not stateful.");var t=this.inputSpec[0].shape[0];if(null==t)throw new p.ValueError("If an RNN is stateful, it needs to know its batch size. Specify the batch size of your input tensors: \n- If using a Sequential model, specify the batch size by passing a `batchInputShape` option to your first layer.\n- If using the functional API, specify the batch size by passing a `batchShape` option to your Input layer.");if(null==this.states)Array.isArray(this.cell.stateSize)?this.states=this.cell.stateSize.map(function(e){return s.zeros([t,e])}):this.states=[s.zeros([t,this.cell.stateSize])];else if(null==e)Array.isArray(this.cell.stateSize)?this.states=this.cell.stateSize.map(function(e){return s.zeros([t,e])}):this.states[0]=s.zeros([t,this.cell.stateSize]);else{if(Array.isArray(e)||(e=[e]),e.length!==this.states.length)throw new p.ValueError("Layer "+this.name+" expects "+this.states.length+" state(s), but it received "+e.length+" state value(s). Input received: "+e);for(var n=0;n<this.states.length;++n){var r=e[n],i=Array.isArray(this.cell.stateSize)?this.cell.stateSize[n]:this.cell.stateSize,o=[t,i];if(!a.util.arraysEqual(r.shape,o))throw new p.ValueError("State "+n+" is incompatible with layer "+this.name+": expected shape="+o+", received shape="+r.shape);this.states[n]=r}}},t.prototype.standardizeArgs=function(e,t,n){function r(e){return null==e||Array.isArray(e)?e:[e]}if(Array.isArray(e)){if(null!=t||null!=n)throw new p.ValueError("When inputs is an array, neither initialState or constants should be provided");null!=this.numConstants&&(n=e.slice(e.length-this.numConstants,e.length),e=e.slice(0,e.length-this.numConstants)),e.length>1&&(t=e.slice(1,e.length)),e=e[0]}return t=r(t),n=r(n),{inputs:e,initialState:t,constants:n}},t.prototype.apply=function(t,n){var r=null==n?null:n.initialState,i=null==n?null:n.constants;null==n&&(n={});var a=this.standardizeArgs(t,r,i);t=a.inputs,r=a.initialState,i=a.constants;var o=[],s=[];if(null!=r){n.initialState=r,o=o.concat(r),this.stateSpec=[];for(var u=0,c=r;u<c.length;u++){var p=c[u];this.stateSpec.push(new l.InputSpec({shape:p.shape}))}s=s.concat(this.stateSpec)}if(null!=i&&(n.constants=i,o=o.concat(i),this.numConstants=i.length),o[0]instanceof f.SymbolicTensor){var h=[t].concat(o),d=this.inputSpec.concat(s),g=this.inputSpec;this.inputSpec=d;var m=e.prototype.apply.call(this,h,n);return this.inputSpec=g,m}return e.prototype.apply.call(this,t,n)},t.prototype.call=function(e,t){var n=this,r=null==t?null:t.mask,i=null==t?null:t.training,a=null==t?null:t.initialState;if(e=g.getExactlyOneTensor(e),null==a){if(this.stateful)throw new p.NotImplementedError("stateful RNN layer is not implemented yet.");a=this.getInitialState(e)}if(null!=r)throw new p.NotImplementedError("Masking is not implemented for RNN yet");var o=Array.isArray(this.cell.stateSize)?this.cell.stateSize.length:1;if(a.length!==o)throw new p.ValueError("RNN Layer has "+o+" state(s) but was passed "+a.length+" initial state(s).");var u=e.shape[1];this.unroll&&console.warn("Ignoring unroll = true for RNN layer, due to imperative backend.");var l={training:i},c=s.rnn(function(e,t){var r=n.cell.call([e].concat(t),l);return[r[0],r.slice(1)]},e,a,this.goBackwards,null,null,this.unroll,u),h=c[0],d=c[1],f=c[2];if(this.stateful)throw new p.NotImplementedError("stateful RNN layer is not implemented yet");var m=this.returnSequences?d:h;return this.returnState?[m].concat(f):m},t.prototype.getInitialState=function(e){var t=s.zeros(e.shape);return t=s.sum(t,[1,2]),t=s.expandDims(t),Array.isArray(this.cell.stateSize)?this.cell.stateSize.map(function(e){return e>1?s.tile(t,[1,e]):t}):this.cell.stateSize>1?[s.tile(t,[1,this.cell.stateSize])]:[t]},Object.defineProperty(t.prototype,"trainableWeights",{get:function(){return this.trainable?this.cell.trainableWeights:[]},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){return this.trainable?this.cell.nonTrainableWeights:this.cell.weights},enumerable:!0,configurable:!0}),t.prototype.getClassName=function(){return"RNN"},t.prototype.getConfig=function(){var t={returnSequences:this.returnSequences,returnState:this.returnState,goBackwards:this.goBackwards,stateful:this.stateful,unroll:this.unroll};null!=this.numConstants&&(t.numConstants=this.numConstants);var n=this.cell.getConfig();t.cell={className:this.cell.getClassName(),config:n};var r=e.prototype.getConfig.call(this);return Object.assign(t,r),t},t}(c.Layer);n.RNN=y,g.ClassNameMap.register("RNN",y);var b=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return r(t,e),t=i([a.doc({heading:"Layers",subheading:"Classes"})],t)}(c.Layer);n.RNNCell=b;var w=function(e){function t(t){var n=e.call(this,t)||this;return n.DEFAULT_ACTIVATION="tanh",n.DEFAULT_KERNEL_INITIALIZER="glorotNormal",n.DEFAULT_RECURRENT_INITIALIZER="orthogonal",n.DEFAULT_BIAS_INITIALIZER="zeros",n.units=t.units,n.activation=o.getActivation(null==t.activation?n.DEFAULT_ACTIVATION:t.activation),n.useBias=null==t.useBias||t.useBias,n.kernelInitializer=h.getInitializer(t.kernelInitializer||n.DEFAULT_KERNEL_INITIALIZER),n.recurrentInitializer=h.getInitializer(t.recurrentInitializer||n.DEFAULT_RECURRENT_INITIALIZER),n.biasInitializer=h.getInitializer(t.biasInitializer||n.DEFAULT_BIAS_INITIALIZER),n.kernelRegularizer=d.getRegularizer(t.kernelRegularizer),n.recurrentRegularizer=d.getRegularizer(t.recurrentRegularizer),n.biasRegularizer=d.getRegularizer(t.biasRegularizer),n.kernelConstraint=u.getConstraint(t.kernelConstraint),n.recurrentConstraint=u.getConstraint(t.recurrentConstraint),n.biasConstraint=u.getConstraint(t.biasConstraint),n.dropout=m.min([1,m.max([0,null==t.dropout?0:t.dropout])]),n.recurrentDropout=m.min([1,m.max([0,null==t.recurrentDropout?0:t.recurrentDropout])]),n.stateSize=n.units,n}return r(t,e),t.prototype.build=function(e){e=g.getExactlyOneShape(e),this.kernel=this.addWeight("kernel",[e[e.length-1],this.units],null,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.recurrentKernel=this.addWeight("recurrent_kernel",[this.units,this.units],null,this.recurrentInitializer,this.recurrentRegularizer,!0,this.recurrentConstraint),this.useBias?this.bias=this.addWeight("bias",[this.units],null,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint):this.bias=null,this.built=!0},t.prototype.call=function(e,t){if(2!==(e=e).length)throw new p.ValueError("SimpleRNNCell expects 2 input Tensors, got "+e.length+".");var n=e[1];if(e=e[0],0!==this.dropout||0!==this.recurrentDropout)throw new p.NotImplementedError("Dropout is not implemented for SimpleRNNCell yet");var r=s.dot(e,this.kernel.read());null!=this.bias&&(r=s.biasAdd(r,this.bias.read()));var i=s.add(r,s.dot(n,this.recurrentKernel.read()));return null!=this.activation&&(i=this.activation(i)),[i,i]},t.prototype.getClassName=function(){return"SimpleRNNCell"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(b);n.SimpleRNNCell=w,g.ClassNameMap.register("SimpleRNNCell",w);var x=function(e){function t(t){return t.cell=new w(t),e.call(this,t)||this}return r(t,e),t.prototype.call=function(t,n){var r=null==n?null:n.mask,i=null==n?null:n.training,a=null==n?null:n.initialState;return e.prototype.call.call(this,t,{mask:r,training:i,initialState:a})},Object.defineProperty(t.prototype,"units",{get:function(){return this.cell.units},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"activation",{get:function(){return this.cell.activation},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"useBias",{get:function(){return this.cell.useBias},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelInitializer",{get:function(){return this.cell.kernelInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentInitializer",{get:function(){return this.cell.recurrentInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasInitializer",{get:function(){return this.cell.biasInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelRegularizer",{get:function(){return this.cell.kernelRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentRegularizer",{get:function(){return this.cell.recurrentRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasRegularizer",{get:function(){return this.cell.biasRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelConstraint",{get:function(){return this.cell.kernelConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentConstraint",{get:function(){return this.cell.recurrentConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasConstraint",{get:function(){return this.cell.biasConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"dropout",{get:function(){return this.cell.dropout},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentDropout",{get:function(){return this.cell.recurrentDropout},enumerable:!0,configurable:!0}),t.prototype.getClassName=function(){return"SimpleRNN"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(y);n.SimpleRNN=x,g.ClassNameMap.register("SimpleRNN",x);var A=function(e){function t(t){var n=e.call(this,t)||this;return n.DEFAULT_ACTIVATION="tanh",n.DEFAULT_RECURRENT_ACTIVATION="hardSigmoid",n.DEFAULT_KERNEL_INITIALIZER="glorotNormal",n.DEFAULT_RECURRENT_INITIALIZER="orthogonal",n.DEFAULT_BIAS_INITIALIZER="zeros",n.units=t.units,n.activation=o.getActivation(void 0===t.activation?n.DEFAULT_ACTIVATION:t.activation),n.recurrentActivation=o.getActivation(void 0===t.activation?n.DEFAULT_RECURRENT_ACTIVATION:t.recurrentActivation),n.useBias=null==t.useBias||t.useBias,n.kernelInitializer=h.getInitializer(t.kernelInitializer||n.DEFAULT_KERNEL_INITIALIZER),n.recurrentInitializer=h.getInitializer(t.recurrentInitializer||n.DEFAULT_RECURRENT_INITIALIZER),n.biasInitializer=h.getInitializer(t.biasInitializer||n.DEFAULT_BIAS_INITIALIZER),n.kernelRegularizer=d.getRegularizer(t.kernelRegularizer),n.recurrentRegularizer=d.getRegularizer(t.recurrentRegularizer),n.biasRegularizer=d.getRegularizer(t.biasRegularizer),n.kernelConstraint=u.getConstraint(t.kernelConstraint),n.recurrentConstraint=u.getConstraint(t.recurrentConstraint),n.biasConstraint=u.getConstraint(t.biasConstraint),n.dropout=m.min([1,m.max([0,null==t.dropout?0:t.dropout])]),n.recurrentDropout=m.min([1,m.max([0,null==t.recurrentDropout?0:t.recurrentDropout])]),n.implementation=t.implementation,n.stateSize=n.units,n}return r(t,e),t.prototype.build=function(e){var t=(e=g.getExactlyOneShape(e))[e.length-1];this.kernel=this.addWeight("kernel",[t,3*this.units],null,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.recurrentKernel=this.addWeight("recurrent_kernel",[this.units,3*this.units],null,this.recurrentInitializer,this.recurrentRegularizer,!0,this.recurrentConstraint),this.useBias?this.bias=this.addWeight("bias",[3*this.units],null,this.biasInitializer,this.biasRegularizer,!0,this.biasConstraint):this.bias=null,this.built=!0},t.prototype.call=function(e,t){if(0!==this.dropout||0!==this.recurrentDropout)throw new p.NotImplementedError("Dropout is not implemented for GRUCell yet");if(2!==(e=e).length)throw new p.ValueError("GRUCell expects 2 input Tensors (inputs, h, c), got "+e.length+".");var n=e[1];e=e[0];var r,i,a;if(1===this.implementation){var o=s.sliceAlongLastAxis(this.kernel.read(),0,this.units),u=s.sliceAlongLastAxis(this.kernel.read(),this.units,this.units),l=s.sliceAlongLastAxis(this.kernel.read(),2*this.units,this.units),c=s.sliceAlongLastAxis(this.recurrentKernel.read(),0,this.units),h=s.sliceAlongLastAxis(this.recurrentKernel.read(),this.units,this.units),d=s.sliceAlongLastAxis(this.recurrentKernel.read(),2*this.units,this.units),f=e,g=e,m=e,v=s.dot(f,o),y=s.dot(g,u),b=s.dot(m,l);if(this.useBias){var w=s.sliceAlongFirstAxis(this.bias.read(),0,this.units),x=s.sliceAlongFirstAxis(this.bias.read(),this.units,this.units),A=s.sliceAlongFirstAxis(this.bias.read(),2*this.units,this.units);v=s.biasAdd(v,w),y=s.biasAdd(y,x),b=s.biasAdd(b,A)}var E=n,_=n,T=n;r=this.recurrentActivation(s.add(v,s.dot(E,c))),i=this.recurrentActivation(s.add(y,s.dot(_,h))),a=this.activation(s.add(b,s.dot(s.multiply(i,T),d)))}else{var S=s.dot(e,this.kernel.read());this.useBias&&(S=s.biasAdd(S,this.bias.read()));var C=s.dot(n,s.sliceAlongLastAxis(this.recurrentKernel.read(),0,2*this.units)),v=s.sliceAlongLastAxis(S,0,this.units),y=s.sliceAlongLastAxis(S,this.units,this.units),O=s.sliceAlongLastAxis(C,0,this.units),k=s.sliceAlongLastAxis(C,this.units,this.units);r=this.recurrentActivation(s.add(v,O)),i=this.recurrentActivation(s.add(y,k));var b=s.sliceAlongLastAxis(S,2*this.units,this.units),R=s.dot(s.multiply(i,n),s.sliceAlongLastAxis(this.recurrentKernel.read(),2*this.units,this.units));a=this.activation(s.add(b,R))}var N=s.add(s.multiply(r,n),s.multiply(s.scalarPlusArray(s.getScalar(1),s.neg(r)),a));return[N,N]},t.prototype.getClassName=function(){return"GRUCell"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout,implementation:this.implementation},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(b);n.GRUCell=A,g.ClassNameMap.register("GRUCell",A);var E=function(e){function t(t){return 0===t.implementation&&console.warn("`implementation=0` has been deprecated, and now defaults to `implementation=1`. Please update your layer call."),t.cell=new A(t),e.call(this,t)||this}return r(t,e),t.prototype.call=function(t,n){var r=null==n?null:n.mask,i=null==n?null:n.training,a=null==n?null:n.initialState;return e.prototype.call.call(this,t,{mask:r,training:i,initialState:a})},Object.defineProperty(t.prototype,"units",{get:function(){return this.cell.units},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"activation",{get:function(){return this.cell.activation},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"useBias",{get:function(){return this.cell.useBias},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelInitializer",{get:function(){return this.cell.kernelInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentInitializer",{get:function(){return this.cell.recurrentInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasInitializer",{get:function(){return this.cell.biasInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelRegularizer",{get:function(){return this.cell.kernelRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentRegularizer",{get:function(){return this.cell.recurrentRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasRegularizer",{get:function(){return this.cell.biasRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelConstraint",{get:function(){return this.cell.kernelConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentConstraint",{get:function(){return this.cell.recurrentConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasConstraint",{get:function(){return this.cell.biasConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"dropout",{get:function(){return this.cell.dropout},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentDropout",{get:function(){return this.cell.recurrentDropout},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"implementation",{get:function(){return this.cell.implementation},enumerable:!0,configurable:!0}),t.prototype.getClassName=function(){return"GRU"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout,implementation:this.implementation},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t.fromConfig=function(e,t){return 0===t.implmentation&&(t.implementation=1),new e(t)},t}(y);n.GRU=E,g.ClassNameMap.register("GRU",E);var _=function(e){function t(t){var n=e.call(this,t)||this;return n.DEFAULT_ACTIVATION="tanh",n.DEFAULT_RECURRENT_ACTIVATION="hardSigmoid",n.DEFAULT_KERNEL_INITIALIZER="glorotNormal",n.DEFAULT_RECURRENT_INITIALIZER="orthogonal",n.DEFAULT_BIAS_INITIALIZER="zeros",n.units=t.units,n.activation=o.getActivation(void 0===t.activation?n.DEFAULT_ACTIVATION:t.activation),n.recurrentActivation=o.getActivation(void 0===t.activation?n.DEFAULT_RECURRENT_ACTIVATION:t.recurrentActivation),n.useBias=null==t.useBias||t.useBias,n.kernelInitializer=h.getInitializer(t.kernelInitializer||n.DEFAULT_KERNEL_INITIALIZER),n.recurrentInitializer=h.getInitializer(t.recurrentInitializer||n.DEFAULT_RECURRENT_INITIALIZER),n.biasInitializer=h.getInitializer(t.biasInitializer||n.DEFAULT_BIAS_INITIALIZER),n.unitForgetBias=t.unitForgetBias,n.kernelRegularizer=d.getRegularizer(t.kernelRegularizer),n.recurrentRegularizer=d.getRegularizer(t.recurrentRegularizer),n.biasRegularizer=d.getRegularizer(t.biasRegularizer),n.kernelConstraint=u.getConstraint(t.kernelConstraint),n.recurrentConstraint=u.getConstraint(t.recurrentConstraint),n.biasConstraint=u.getConstraint(t.biasConstraint),n.dropout=m.min([1,m.max([0,null==t.dropout?0:t.dropout])]),n.recurrentDropout=m.min([1,m.max([0,null==t.recurrentDropout?0:t.recurrentDropout])]),n.implementation=t.implementation,n.stateSize=[n.units,n.units],n}return r(t,e),t.prototype.build=function(e){var t=(e=g.getExactlyOneShape(e))[e.length-1];this.kernel=this.addWeight("kernel",[t,4*this.units],null,this.kernelInitializer,this.kernelRegularizer,!0,this.kernelConstraint),this.recurrentKernel=this.addWeight("recurrent_kernel",[this.units,4*this.units],null,this.recurrentInitializer,this.recurrentRegularizer,!0,this.recurrentConstraint);var n;if(this.useBias){if(this.unitForgetBias){var i=this.biasInitializer,a=this.units;n=new(function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return r(t,e),t.prototype.apply=function(e,t){var n=i.apply([a]),r=(new h.Ones).apply([a]),o=i.apply([2*a]);return s.concatAlongFirstAxis(s.concatAlongFirstAxis(n,r),o)},t.prototype.getClassName=function(){return"CustomInit"},t}(h.Initializer))}else n=this.biasInitializer;this.bias=this.addWeight("bias",[4*this.units],null,n,this.biasRegularizer,!0,this.biasConstraint)}else this.bias=null;this.built=!0},t.prototype.call=function(e,t){if(0!==this.dropout||0!==this.recurrentDropout)throw new p.NotImplementedError("Dropout is not implemented for LSTMCell yet");if(3!==(e=e).length)throw new p.ValueError("LSTMCell expects 3 input Tensors (inputs, h, c), got "+e.length+".");var n=e[1],r=e[2];e=e[0];var i,a,o,u;if(1===this.implementation){var l=s.sliceAlongLastAxis(this.kernel.read(),0,this.units),c=s.sliceAlongLastAxis(this.kernel.read(),this.units,this.units),h=s.sliceAlongLastAxis(this.kernel.read(),2*this.units,this.units),d=s.sliceAlongLastAxis(this.kernel.read(),3*this.units,this.units),f=s.sliceAlongLastAxis(this.recurrentKernel.read(),0,this.units),g=s.sliceAlongLastAxis(this.recurrentKernel.read(),this.units,this.units),m=s.sliceAlongLastAxis(this.recurrentKernel.read(),2*this.units,this.units),v=s.sliceAlongLastAxis(this.recurrentKernel.read(),3*this.units,this.units),y=e,b=e,w=e,x=e,A=s.dot(y,l),E=s.dot(b,c),_=s.dot(w,h),T=s.dot(x,d);if(this.useBias){var S=s.sliceAlongFirstAxis(this.bias.read(),0,this.units),C=s.sliceAlongFirstAxis(this.bias.read(),this.units,this.units),O=s.sliceAlongFirstAxis(this.bias.read(),2*this.units,this.units),k=s.sliceAlongFirstAxis(this.bias.read(),3*this.units,this.units);A=s.biasAdd(A,S),E=s.biasAdd(E,C),_=s.biasAdd(_,O),T=s.biasAdd(T,k)}var R=n,N=n,I=n,z=n;i=this.recurrentActivation(s.add(A,s.dot(R,f))),a=this.recurrentActivation(s.add(E,s.dot(N,g))),o=s.add(s.multiply(a,r),s.multiply(i,this.activation(s.add(_,s.dot(I,m))))),u=this.recurrentActivation(s.add(T,s.dot(z,v)))}else{var M=s.dot(e,this.kernel.read());M=s.add(M,s.dot(n,this.recurrentKernel.read())),this.useBias&&(M=s.biasAdd(M,this.bias.read()));var D=s.sliceAlongLastAxis(M,0,this.units),P=s.sliceAlongLastAxis(M,this.units,this.units),L=s.sliceAlongLastAxis(M,2*this.units,this.units),F=s.sliceAlongLastAxis(M,3*this.units,this.units);i=this.recurrentActivation(D),a=this.recurrentActivation(P),o=s.add(s.multiply(a,r),s.multiply(i,this.activation(L))),u=this.recurrentActivation(F)}var V=s.multiply(u,this.activation(o));return[V,V,o]},t.prototype.getClassName=function(){return"LSTMCell"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),unitForgetBias:this.unitForgetBias,kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout,implementation:this.implementation},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t}(b);n.LSTMCell=_,g.ClassNameMap.register("LSTMCell",_);var T=function(e){function t(t){return 0===t.implementation&&console.warn("`implementation=0` has been deprecated, and now defaults to `implementation=1`. Please update your layer call."),t.cell=new _(t),e.call(this,t)||this}return r(t,e),t.prototype.call=function(t,n){var r=null==n?null:n.mask,i=null==n?null:n.training,a=null==n?null:n.initialState;return e.prototype.call.call(this,t,{mask:r,training:i,initialState:a})},Object.defineProperty(t.prototype,"units",{get:function(){return this.cell.units},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"activation",{get:function(){return this.cell.activation},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"useBias",{get:function(){return this.cell.useBias},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelInitializer",{get:function(){return this.cell.kernelInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentInitializer",{get:function(){return this.cell.recurrentInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasInitializer",{get:function(){return this.cell.biasInitializer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"unitForgetBias",{get:function(){return this.cell.unitForgetBias},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelRegularizer",{get:function(){return this.cell.kernelRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentRegularizer",{get:function(){return this.cell.recurrentRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasRegularizer",{get:function(){return this.cell.biasRegularizer},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"kernelConstraint",{get:function(){return this.cell.kernelConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentConstraint",{get:function(){return this.cell.recurrentConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"biasConstraint",{get:function(){return this.cell.biasConstraint},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"dropout",{get:function(){return this.cell.dropout},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"recurrentDropout",{get:function(){return this.cell.recurrentDropout},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"implementation",{get:function(){return this.cell.implementation},enumerable:!0,configurable:!0}),t.prototype.getClassName=function(){return"LSTM"},t.prototype.getConfig=function(){var t={units:this.units,activation:o.serializeActivation(this.activation),useBias:this.useBias,kernelInitializer:h.serializeInitializer(this.kernelInitializer),recurrentInitializer:h.serializeInitializer(this.recurrentInitializer),biasInitializer:h.serializeInitializer(this.biasInitializer),unitForgetBias:this.unitForgetBias,kernelRegularizer:d.serializeRegularizer(this.kernelRegularizer),recurrentRegularizer:d.serializeRegularizer(this.recurrentRegularizer),biasRegularizer:d.serializeRegularizer(this.biasRegularizer),activityRegularizer:d.serializeRegularizer(this.activityRegularizer),kernelConstraint:u.serializeConstraint(this.kernelConstraint),recurrentConstraint:u.serializeConstraint(this.recurrentConstraint),biasConstraint:u.serializeConstraint(this.biasConstraint),dropout:this.dropout,recurrentDropout:this.recurrentDropout,implementation:this.implementation},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t.fromConfig=function(e,t){return 0===t.implmentation&&(t.implementation=1),new e(t)},t}(y);n.LSTM=T,g.ClassNameMap.register("LSTM",T);var S=function(e){function t(t){var n=e.call(this,t)||this;return n.cells=t.cells,n}return r(t,e),Object.defineProperty(t.prototype,"stateSize",{get:function(){for(var e=[],t=0,n=this.cells.slice().reverse();t<n.length;t++){var r=n[t];Array.isArray(r.stateSize)?e.push.apply(e,r.stateSize):e.push(r.stateSize)}return e},enumerable:!0,configurable:!0}),t.prototype.call=function(e,t){for(var n=(e=e).slice(1),r=[],i=0,a=this.cells.slice().reverse();i<a.length;i++){l=a[i];Array.isArray(l.stateSize)?r.push(n.splice(0,l.stateSize.length)):r.push(n.splice(0,1))}r.reverse();for(var o,s=[],u=0;u<this.cells.length;++u){var l=this.cells[u];n=r[u],o=0===u?[e[0]].concat(n):[o[0]].concat(n),o=l.call(o,t),s.push(o.slice(1))}n=[];for(var c=0,p=s.slice().reverse();c<p.length;c++){var h=p[c];n.push.apply(n,h)}return[o[0]].concat(n)},t.prototype.build=function(e){g.isArrayOfShapes(e)&&(e=e[0]),e=e;for(var t,n=0,r=this.cells;n<r.length;n++){var i=r[n];i.build(e),t=Array.isArray(i.stateSize)?i.stateSize[0]:i.stateSize,e=[e[0],t]}this.built=!0},t.prototype.getClassName=function(){return"StackedRNNCells"},t.prototype.getConfig=function(){for(var t=[],n=0,r=this.cells;n<r.length;n++){var i=r[n];t.push({className:this.getClassName(),config:i.getConfig()})}var a={cells:t},o=e.prototype.getConfig.call(this);return Object.assign(a,o),a},t.fromConfig=function(e,t,n){void 0===n&&(n={});for(var r=[],i=0,a=t.cells;i<a.length;i++){var o=a[i];r.push(v.deserialize(o,n))}return new e({cells:r})},Object.defineProperty(t.prototype,"trainableWeights",{get:function(){if(!this.trainable)return[];for(var e=[],t=0,n=this.cells;t<n.length;t++){var r=n[t];e.push.apply(e,r.trainableWeights)}return e},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){for(var e=[],t=0,n=this.cells;t<n.length;t++){o=n[t];e.push.apply(e,o.nonTrainableWeights)}if(!this.trainable){for(var r=[],i=0,a=this.cells;i<a.length;i++){var o=a[i];r.push.apply(r,o.trainableWeights)}return r.concat(e)}return e},enumerable:!0,configurable:!0}),t.prototype.getWeights=function(){for(var e=[],t=0,n=this.cells;t<n.length;t++){var r=n[t];e.push.apply(e,r.weights)}return s.batchGetValue(e)},t.prototype.setWeights=function(e){for(var t=[],n=0,r=this.cells;n<r.length;n++)for(var i=r[n],a=i.weights.length,o=e.splice(a),u=0;u<i.weights.length;++u)t.push([i.weights[u],o[u]]);s.batchSetValue(t)},t}(b);n.StackedRNNCells=S,g.ClassNameMap.register("StackedRNNCells",S)},{"../activations":98,"../backend/tfjs_backend":100,"../constraints":103,"../engine/topology":105,"../errors":107,"../initializers":110,"../regularizers":127,"../types":128,"../utils/generic_utils":130,"../utils/math_utils":131,"./serialization":121,"@tensorflow/tfjs-core":8}],121:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../utils/generic_utils");n.deserialize=function(e,t){return void 0===t&&(t={}),r.deserializeKerasObject(e,r.ClassNameMap.getMap().pythonClassNameMap,t,"layer")}},{"../utils/generic_utils":130}],122:[function(e,t,n){"use strict";var r=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}();Object.defineProperty(n,"__esModule",{value:!0});var i=e("../backend/tfjs_backend"),a=e("../engine/topology"),o=e("../errors"),s=e("../utils/generic_utils"),u=e("./serialization"),l=function(e){function t(t){var n=e.call(this,t)||this;return n.layer=t.layer,n}return r(t,e),t.prototype.build=function(e){this.built=!0},Object.defineProperty(t.prototype,"trainable",{get:function(){return null!=this.layer&&this.layer.trainable},set:function(e){null!=this.layer&&(this.layer.trainable=e)},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"trainableWeights",{get:function(){return this.layer.trainableWeights},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){return this.layer.nonTrainableWeights},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"updates",{get:function(){return this.layer._updates},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"losses",{get:function(){return this.layer.losses},enumerable:!0,configurable:!0}),t.prototype.getWeights=function(){return this.layer.getWeights()},t.prototype.setWeights=function(e){this.layer.setWeights(e)},t.prototype.getConfig=function(){var t={layer:{className:this.layer.getClassName(),config:this.layer.getConfig()}},n=e.prototype.getConfig.call(this);return Object.assign(t,n),t},t.fromConfig=function(e,t,n){void 0===n&&(n={});var r=t.layer,i=u.deserialize(r,n);delete t.layer;var a={layer:i};return Object.assign(a,t),new e(a)},t}(a.Layer);n.Wrapper=l;var c=function(e){function t(t){var n=e.call(this,t)||this;return n.supportsMasking=!0,n}return r(t,e),t.prototype.build=function(t){if((t=s.getExactlyOneShape(t)).length<3)throw new o.ValueError("TimeDistributed layer expects an input shape >= 3D, but received input shape "+JSON.stringify(t));this.inputSpec=[{shape:t}];var n=[t[0]].concat(t.slice(2));this.layer.built||(this.layer.build(n),this.layer.built=!0),e.prototype.build.call(this,t)},t.prototype.computeOutputShape=function(e){var t=[(e=s.getExactlyOneShape(e))[0]].concat(e.slice(2)),n=this.layer.computeOutputShape(t),r=e[1];return[n[0],r].concat(n.slice(1))},t.prototype.call=function(e,t){var n=this;e=s.getExactlyOneTensor(e);return i.rnn(function(e,r){return[n.layer.call(e,t),[]]},e,[],!1,null,null,!1,e.shape[1])[1]},t.prototype.getClassName=function(){return"TimeDistributed"},t}(l);n.TimeDistributed=c,s.ClassNameMap.register("TimeDistributed",c);var p;!function(e){e[e.SUM=0]="SUM",e[e.MUL=1]="MUL",e[e.CONCAT=2]="CONCAT",e[e.AVE=3]="AVE"}(p=n.BidirectionalMergeMode||(n.BidirectionalMergeMode={})),s.SerializableEnumRegistry.register("merge_mode",{sum:p.SUM,mul:p.MUL,concat:p.CONCAT,ave:p.AVE});var h=function(e){function t(t){var n=e.call(this,t)||this;n.forwardLayer=t.layer;var r=t.layer.getConfig();if(r.goBackwards=!0!==r.goBackwards,n.backwardLayer=u.deserialize({className:t.layer.getClassName(),config:r}),n.forwardLayer.name="forward_"+n.forwardLayer.name,n.backwardLayer.name="backward_"+n.backwardLayer.name,n.mergeMode=t.mergeMode,t.weights)throw new o.NotImplementedError("weights support is not implemented for Bidirectional layer yet.");return n._stateful=t.layer.stateful,n.returnSequences=t.layer.returnSequences,n.returnState=t.layer.returnState,n.supportsMasking=!0,n._trainable=!0,n.inputSpec=t.layer.inputSpec,n}return r(t,e),Object.defineProperty(t.prototype,"trainable",{get:function(){return this._trainable},set:function(e){this._trainable=e,null!=this.forwardLayer&&(this.forwardLayer.trainable=e),null!=this.backwardLayer&&(this.backwardLayer.trainable=e)},enumerable:!0,configurable:!0}),t.prototype.getWeights=function(){return this.forwardLayer.getWeights().concat(this.backwardLayer.getWeights())},t.prototype.setWeights=function(e){var t=e.length,n=Math.floor(t/2);this.forwardLayer.setWeights(e.slice(0,n)),this.backwardLayer.setWeights(e.slice(n))},t.prototype.computeOutputShape=function(e){var t=this.forwardLayer.computeOutputShape(e);Array.isArray(t)&&Array.isArray(t[0])||(t=[t]),t=t;var n,r,i;return this.returnState?(i=t.slice(1),n=t[0]):n=t[0],n=n,this.mergeMode===p.CONCAT?(n[n.length-1]*=2,r=[n]):r=null==this.mergeMode?[n,n.slice()]:[n],this.returnState?null==this.mergeMode?r.concat(i).concat(i.slice()):[n].concat(i).concat(i.slice()):s.singletonOrArray(r)},t.prototype.apply=function(t,n){var r=null;if(null!=n&&(r=n.initialState),Array.isArray(t)&&(r=t.slice(1),t=t[0]),null==r||0===r.length)return e.prototype.apply.call(this,t,n);throw new o.NotImplementedError("The support for initial states is not implemented for Bidirectional layers yet.")},t.prototype.call=function(e,t){if(null!=t.mask)throw new o.NotImplementedError("The support for masking is not implemented for Bidirectional layers yet.");if(null!=t.initialState)throw new o.NotImplementedError("The support for initial states is not implemented for Bidirectional layers yet.");var n,r=this.forwardLayer.call(e,t),a=this.backwardLayer.call(e,t);this.returnState&&(Array.isArray(r)&&(n=r.slice(1).concat(a.slice(1))),r=r[0],a=a[0]),this.returnSequences&&(a=i.reverse(a,1));var s;return this.mergeMode===p.CONCAT?s=i.concatenate([r,a]):this.mergeMode===p.SUM?s=i.add(r,a):this.mergeMode===p.AVE?s=i.scalarTimesArray(i.getScalar(.5),i.add(r,a)):this.mergeMode===p.MUL?s=i.multiply(r,a):null==this.mergeMode&&(s=[r,a]),this.returnState?null==this.mergeMode?s.concat(n):[s].concat(n):s},t.prototype.resetStates=function(e){this.forwardLayer.resetStates(),this.backwardLayer.resetStates()},t.prototype.build=function(e){var t=this;i.nameScope(this.forwardLayer.name,function(){t.forwardLayer.build(e)}),i.nameScope(this.backwardLayer.name,function(){t.backwardLayer.build(e)}),this.built=!0},Object.defineProperty(t.prototype,"trainableWeights",{get:function(){return this.forwardLayer.trainableWeights.concat(this.backwardLayer.trainableWeights)},enumerable:!0,configurable:!0}),Object.defineProperty(t.prototype,"nonTrainableWeights",{get:function(){return this.forwardLayer.nonTrainableWeights.concat(this.backwardLayer.nonTrainableWeights)},enumerable:!0,configurable:!0}),t.prototype.getClassName=function(){return"Bidirectional"},t}(l);n.Bidirectional=h,s.ClassNameMap.register("Bidirectional",h)},{"../backend/tfjs_backend":100,"../engine/topology":105,"../errors":107,"../utils/generic_utils":130,"./serialization":121}],123:[function(e,t,n){"use strict";function r(e,t){return v.mean(v.square(v.subtract(t,e)),-1)}function i(e,t){return v.mean(v.abs(v.subtract(t,e)),-1)}function a(e,t){var n=v.subtract(e,t),r=v.clip(v.abs(e),v.epsilon(),Number.MAX_VALUE),i=v.abs(v.divide(n,r));return v.scalarTimesArray(v.getScalar(100),v.mean(i,-1))}function o(e,t){var n=v.getScalar(1),r=v.clip(t,v.epsilon(),Number.MAX_VALUE),i=v.log(v.scalarPlusArray(n,r)),a=v.clip(e,v.epsilon(),Number.MAX_VALUE),o=v.log(v.scalarPlusArray(n,a));return v.mean(v.square(v.subtract(i,o)),-1)}function s(e,t){var n=v.getScalar(0),r=v.getScalar(1),i=v.maximum(n,v.subtract(r,v.multiply(e,t)));return v.mean(v.square(i),-1)}function u(e,t){var n=v.getScalar(0),r=v.getScalar(1),i=v.maximum(n,v.subtract(r,v.multiply(e,t)));return v.mean(i,-1)}function l(e,t){var n=v.getScalar(0),r=v.getScalar(1),i=v.sum(v.multiply(e,t),-1),a=v.max(v.multiply(v.subtract(r,e),t),-1);return v.maximum(n,v.scalarPlusArray(r,v.subtract(a,i)))}function c(e,t){var n=v.getScalar(Math.log(2)),r=v.subtract(t,e),i=v.subtract(v.add(r,v.softplus(v.scalarTimesArray(v.getScalar(-2),r))),n);return v.mean(i,-1)}function p(e,t){return v.categoricalCrossentropy(e,t)}function h(e,t){return v.sparseCategoricalCrossentropy(e,t)}function d(e,t){return v.mean(v.binaryCrossentropy(e,t),-1)}function f(e,t){var n=v.clip(e,v.epsilon(),1),r=v.clip(t,v.epsilon(),1);return v.sum(v.multiply(e,v.log(v.divide(n,r))),-1)}function g(e,t){var n=v.log(v.scalarPlusArray(v.getScalar(v.epsilon()),t));return v.mean(v.subtract(t,v.multiply(e,n)),-1)}function m(e,t){var n=v.l2Normalize(e,-1),r=v.l2Normalize(t,-1),i=v.multiply(n,r);return v.neg(v.sum(i,-1))}Object.defineProperty(n,"__esModule",{value:!0});var v=e("./backend/tfjs_backend"),y=e("./errors");n.meanSquaredError=r,n.meanAbsoluteError=i,n.meanAbsolutePercentageError=a,n.meanSquaredLogarithmicError=o,n.squaredHinge=s,n.hinge=u,n.categoricalHinge=l,n.logcosh=c,n.categoricalCrossentropy=p,n.sparseCategoricalCrossentropy=h,n.binaryCrossentropy=d,n.kullbackLeiblerDivergence=f,n.poisson=g,n.cosineProximity=m,n.mse=r,n.MSE=r,n.mae=i,n.MAE=i,n.mape=a,n.MAPE=a,n.msle=o,n.MSLE=o,n.kld=f,n.KLD=f,n.cosine=m,n.get=function(e){var t={meanSquaredError:r,meanAbsoluteError:i,meanAbsolutePercentageError:a,meanSquaredLogarithmicError:o,squaredHinge:s,hinge:u,categoricalHinge:l,logcosh:c,categoricalCrossentropy:p,sparseCategoricalCrossentropy:h,binaryCrossentropy:d,kullbackLeiblerDivergence:f,poisson:g,cosineProximity:m};if("string"==typeof e){if(e in t)return t[e];throw new y.ValueError("Unknown loss "+e)}return e}},{"./backend/tfjs_backend":100,"./errors":107}],124:[function(e,t,n){"use strict";function r(e,t){var n=o.scalarTimesArray(o.getScalar(.5),a.onesLike(t)),r=o.cast(o.greater(t,n),e.dtype);return o.mean(o.equal(e,r),-1)}function i(e,t){return o.cast(o.equal(o.argmax(e,-1),o.argmax(t,-1)),"float32")}Object.defineProperty(n,"__esModule",{value:!0});var a=e("@tensorflow/tfjs-core"),o=e("./backend/tfjs_backend"),s=e("./errors"),u=e("./losses");n.binaryAccuracy=r,n.categoricalAccuracy=i,n.binaryCrossentropy=function(e,t){return o.mean(o.binaryCrossentropy(e,t),-1)},n.sparseCategoricalAccuracy=function(e,t){throw new s.NotImplementedError},n.topKCategoricalAccuracy=function(e,t){throw new s.NotImplementedError},n.sparseTopKCategoricalAccuracy=function(e,t){throw new s.NotImplementedError},n.mse=u.meanSquaredError,n.MSE=u.meanSquaredError,n.mae=u.meanAbsoluteError,n.MAE=u.meanAbsoluteError,n.mape=u.meanAbsolutePercentageError,n.MAPE=u.meanAbsolutePercentageError,n.categoricalCrossentropy=u.categoricalCrossentropy,n.cosine=u.cosineProximity,n.sparseCategoricalCrossentropy=u.sparseCategoricalCrossentropy,n.get=function(e){var t={binaryAccuracy:r,categoricalAccuracy:i,categoricalCrossentropy:n.categoricalCrossentropy,sparseCategoricalCrossentropy:n.sparseCategoricalCrossentropy,mse:n.mse,MSE:n.MSE,mae:n.mae,MAE:n.MAE,mape:n.mape,MAPE:n.MAPE,cosine:n.cosine};if("string"==typeof e&&e in t)return t[e];if("string"!=typeof e&&null!=e)return e;throw new s.ValueError("Unknown metric "+e)}},{"./backend/tfjs_backend":100,"./errors":107,"./losses":123,"@tensorflow/tfjs-core":8}],125:[function(e,t,n){"use strict";function r(e,t){return o(this,void 0,void 0,function(){var n,r,i,a,o,l,c,p,h,f;return s(this,function(s){switch(s.label){case 0:return null!=(n=e.modelTopology).model_config&&(n=n.model_config),r=g.convertPythonicToTs(n),i=d.deserialize(r,t),null==e.weightsManifest?[3,2]:[4,u.loadWeights(e.weightsManifest,e.pathPrefix,i.weights.map(function(e){return e.originalName}))];case 1:for(a=s.sent(),o={},l=0,c=i.weights;l<c.length;l++)p=c[l],o[p.name]=a[p.originalName];h=null,f=!0,i.loadWeights(o,h,f),s.label=2;case 2:return[2,i]}})})}var i=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),a=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o},o=this&&this.__awaiter||function(e,t,n,r){return new(n||(n=Promise))(function(i,a){function o(e){try{u(r.next(e))}catch(e){a(e)}}function s(e){try{u(r.throw(e))}catch(e){a(e)}}function u(e){e.done?i(e.value):new n(function(t){t(e.value)}).then(o,s)}u((r=r.apply(e,t||[])).next())})},s=this&&this.__generator||function(e,t){function n(e){return function(t){return r([e,t])}}function r(n){if(i)throw new TypeError("Generator is already executing.");for(;u;)try{if(i=1,a&&(o=a[2&n[0]?"return":n[0]?"throw":"next"])&&!(o=o.call(a,n[1])).done)return o;switch(a=0,o&&(n=[0,o.value]),n[0]){case 0:case 1:o=n;break;case 4:return u.label++,{value:n[1],done:!1};case 5:u.label++,a=n[1],n=[0];continue;case 7:n=u.ops.pop(),u.trys.pop();continue;default:if(o=u.trys,!(o=o.length>0&&o[o.length-1])&&(6===n[0]||2===n[0])){u=0;continue}if(3===n[0]&&(!o||n[1]>o[0]&&n[1]<o[3])){u.label=n[1];break}if(6===n[0]&&u.label<o[1]){u.label=o[1],o=n;break}if(o&&u.label<o[2]){u.label=o[2],u.ops.push(n);break}o[2]&&u.ops.pop(),u.trys.pop();continue}n=t.call(e,u)}catch(e){n=[6,e],a=0}finally{i=o=0}if(5&n[0])throw n[1];return{value:n[0]?n[1]:void 0,done:!0}}var i,a,o,s,u={label:0,sent:function(){if(1&o[0])throw o[1];return o[1]},trys:[],ops:[]};return s={next:n(0),throw:n(1),return:n(2)},"function"==typeof Symbol&&(s[Symbol.iterator]=function(){return this}),s};Object.defineProperty(n,"__esModule",{value:!0});var u=e("@tensorflow/tfjs-core"),l=e("./backend/tfjs_backend"),c=e("./engine/topology"),p=e("./engine/training"),h=e("./errors"),d=e("./layers/serialization"),f=e("./utils/generic_utils"),g=e("./utils/serialization_utils");n.modelFromJSON=r,n.loadModelInternal=function(e){return o(this,void 0,void 0,function(){var t,n;return s(this,function(i){switch(i.label){case 0:return[4,fetch(e)];case 1:return t=i.sent(),[4,t.json()];case 2:if(null==(n=i.sent()).modelTopology)throw new h.ValueError('Missing field "modelTopology" from model JSON at path'+e);if(null==n.weightsManifest)throw new h.ValueError('Missing field "weightsManifest" from model JSON at path'+e);return n.pathPrefix=e.substring(0,e.lastIndexOf("/")),[2,r(n)]}})})};var m=function(e){function t(t){var n=e.call(this,{inputs:[],outputs:[]})||this;if(t=t||{},n.trainable=!0,n._updatable=!0,n.built=!1,n.name=null!=t.name?t.name:l.getUid("sequential_"),null!=t.layers)for(var r=0,i=t.layers;r<i.length;r++){var a=i[r];n.add(a)}return n}return i(t,e),n=t,t.prototype.getClassName=function(){return"Sequential"},t.prototype.add=function(e){if(0===this.outputs.length){if(0===e.inboundNodes.length){if(null==e.batchInputShape)throw new h.ValueError("The first layer in a Sequential model must get an `inputShape` or `batchInputShape` argument.");var t=c.Input({batchShape:e.batchInputShape,dtype:e.dtype,name:e.name+"_input"});e.apply(t)}if(1!==e.inboundNodes.length)throw new h.ValueError("A layer added to a Sequential model must not already be connected somewhere else. Model received layer "+e.name+" which has "+e.inboundNodes.length+" pre-existing inbound connections.");if(1!==e.inboundNodes[0].outputTensors.length)throw new h.ValueError("All layers in a Sequential model should have a single output tensor. For multi-output layers, use the functional API.");this.outputs=[e.inboundNodes[0].outputTensors[0]],this.inputs=c.getSourceInputs(this.outputs[0]),new c.Node({outboundLayer:this,inboundLayers:[],nodeIndices:[],tensorIndices:[],inputTensors:this.inputs,outputTensors:this.outputs,inputMasks:f.pyListRepeat(null,this.inputs.length),outputMasks:[null],inputShapes:this.inputs.map(function(e){return e.shape}),outputShapes:this.outputs[0].shape})}else{var n=e.apply(this.outputs[0]);if(Array.isArray(n))throw new TypeError("All layers in a Sequential model should have a single output tensor. For multi-output layers, use the functional API.");this.outputs=[n],this.inboundNodes[0].outputTensors=this.outputs,this.inboundNodes[0].outputShapes=[this.outputs[0].shape]}this.layers.push(e),this.built=!1},t.prototype.pop=function(){if(0===this.layers.length)throw new TypeError("There are no layers in the model.");if(this.layers.pop(),0===this.layers.length)this.outputs=[],this.inboundNodes=[],this.outboundNodes=[];else{var e=this.layers.length-1;this.layers[e].outboundNodes=[],this.outputs=[this.layers[e].output],this.inboundNodes[0].outputTensors=this.outputs,this.inboundNodes[0].outputShapes=[this.outputs[0].shape]}},t.prototype.call=function(e,t){return null==this.model&&this.build(),this.model.call(e,t)},t.prototype.build=function(e){if(0===this.inputs.length||0===this.outputs.length)throw new TypeError("Sequential model cannot be built: model is empty. Add some layers first.");this.model=new p.Model({inputs:this.inputs,outputs:this.outputs[0],name:this.name+"_model"}),this.model.trainable=this.trainable,this.model.updatable=this.updatable,this.supportsMasking=this.model.supportsMasking,this.inputLayers=this.model.inputLayers,this.inputLayersNodeIndices=this.model.inputLayersNodeIndices,this.inputLayersTensorIndices=this.model.inputLayersTensorIndices,this.outputLayers=this.model.outputLayers,this.outputLayersNodeIndices=this.model.outputLayersNodeIndices,this.outputLayersTensorIndices=this.model.outputLayersTensorIndices,this.nodesByDepth=this.model.nodesByDepth,this.containerNodes=this.model.containerNodes,this.outputNames=this.model.outputNames,this.inputNames=this.model.inputNames,this.built=!0},t.prototype.setWeights=function(e){null==this.model&&this.build(),this.model.setWeights(e)},Object.defineProperty(t.prototype,"updatable",{get:function(){return this._updatable},set:function(e){this.built&&(this.model.updatable=e),this._updatable=e},enumerable:!0,configurable:!0}),t.prototype.evaluate=function(e,t,n){if(void 0===n&&(n={}),!this.built)throw new h.RuntimeError("The model needs to be compiled before being used.");return this.model.evaluate(e,t,n)},t.prototype.predict=function(e,t){return void 0===t&&(t={}),null==this.model&&this.build(),this.model.predict(e,t)},t.prototype.predictOnBatch=function(e){return null==this.model&&this.build(),this.model.predictOnBatch(e)},t.prototype.compile=function(e){this.build(),this.model.compile(e),this.optimizer=this.model.optimizer,this.loss=this.model.loss,this.metrics=this.model.metrics,this.metricsTensors=this.model.metricsTensors,this.metricsNames=this.model.metricsNames},t.prototype.fit=function(e,t,n){return void 0===n&&(n={}),o(this,void 0,void 0,function(){return s(this,function(r){if(!this.built)throw new h.RuntimeError("The model needs to be compiled before being used.");return[2,this.model.fit(e,t,n)]})})},t.fromConfig=function(e,t){var r=new e({});if(!(r instanceof n))throw new h.ValueError("Sequential.fromConfig called on non-Sequential input: "+r);if(!(t instanceof Array))throw new h.ValueError("Sequential.fromConfig called without an array of configs");if(null==t[0].className||"Merge"===t[0].className)throw new h.ValueError("Legacy serialization format not supported yet.");for(var i=0,a=t;i<a.length;i++){var o=a[i],s=d.deserialize(o);r.add(s)}return r},t.prototype.getConfig=function(){for(var e=[],t=0,n=this.layers;t<n.length;t++){var r=n[t];e.push({className:r.getClassName(),config:r.getConfig()})}return e},a([u.doc({heading:"Models",subheading:"Classes"})],t.prototype,"add",null),a([u.doc({heading:"Models",subheading:"Classes",configParamIndices:[2]})],t.prototype,"evaluate",null),a([u.doc({heading:"Models",subheading:"Classes",configParamIndices:[1]})],t.prototype,"predict",null),a([u.doc({heading:"Models",subheading:"Classes",configParamIndices:[2]})],t.prototype,"fit",null),t=n=a([u.doc({heading:"Models",subheading:"Classes"})],t);var n}(p.Model);n.Sequential=m,f.ClassNameMap.register("Sequential",m)},{"./backend/tfjs_backend":100,"./engine/topology":105,"./engine/training":106,"./errors":107,"./layers/serialization":121,"./utils/generic_utils":130,"./utils/serialization_utils":132,"@tensorflow/tfjs-core":8}],126:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("@tensorflow/tfjs-core"),i=e("./backend/tfjs_backend"),a=e("./errors");n.getOptimizer=function(e){var t={Adagrad:function(){return r.train.adagrad(.01)},Adam:function(){return r.train.adam(.001,.9,.999,i.epsilon())},RMSProp:function(){return r.train.rmsprop(.001,.9,null,i.epsilon())},SGD:function(){return r.train.sgd(.01)}};if(t.adagrad=t.Adagrad,t.adam=t.Adam,t.rmsprop=t.RMSProp,t.sgd=t.SGD,e in t)return t[e]();throw new a.ValueError("Unknown Optimizer "+e)}},{"./backend/tfjs_backend":100,"./errors":107,"@tensorflow/tfjs-core":8}],127:[function(e,t,n){"use strict";function r(e,t){return void 0===t&&(t={}),l.deserializeKerasObject(e,l.ClassNameMap.getMap().pythonClassNameMap,t,"regularizer")}var i=this&&this.__extends||function(){var e=Object.setPrototypeOf||{__proto__:[]}instanceof Array&&function(e,t){e.__proto__=t}||function(e,t){for(var n in t)t.hasOwnProperty(n)&&(e[n]=t[n])};return function(t,n){function r(){this.constructor=t}e(t,n),t.prototype=null===n?Object.create(n):(r.prototype=n.prototype,new r)}}(),a=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var o=e("@tensorflow/tfjs-core"),s=e("./backend/tfjs_backend"),u=e("./types"),l=e("./utils/generic_utils"),c=function(e){function t(){return null!==e&&e.apply(this,arguments)||this}return i(t,e),t}(u.Serializable);n.Regularizer=c;var p=function(e){function t(t){var n=e.call(this)||this,r=null==t||null==t.l1?.01:t.l1,i=null==t||null==t.l2?.01:t.l2;return n.hasL1=0!==r,n.hasL2=0!==i,n.l1=s.getScalar(r),n.l2=s.getScalar(i),n}return i(t,e),n=t,t.prototype.apply=function(e){var t=o.zeros([1]);return this.hasL1&&(t=s.add(t,s.sum(s.scalarTimesArray(this.l1,s.abs(e))))),this.hasL2&&(t=s.add(t,s.sum(s.scalarTimesArray(this.l2,s.square(e))))),t.asScalar()},t.prototype.getClassName=function(){return"L1L2"},t.prototype.getConfig=function(){return{l1:this.l1.dataSync()[0],l2:this.l2.dataSync()[0]}},t.fromConfig=function(e,t){return new n({l1:t.l1,l2:t.l2})},t=n=a([o.doc({heading:"Regularizers",namespace:"regularizers"})],t);var n}(c);n.L1L2=p,l.ClassNameMap.register("L1L2",p),n.l1=function(e){return new p({l1:null!=e?e.l1:null,l2:0})},n.l2=function(e){return new p({l2:null!=e?e.l2:null,l1:0})},n.REGULARIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP={l1l2:"L1L2"},n.serializeRegularizer=function(e){return l.serializeKerasObject(e)},n.deserializeRegularizer=r,n.getRegularizer=function(e){return null==e?null:"string"==typeof e?r({className:e in n.REGULARIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP?n.REGULARIZER_IDENTIFIER_REGISTRY_SYMBOL_MAP[e]:e,config:{}}):e instanceof c?e:r(e)}},{"./backend/tfjs_backend":100,"./types":128,"./utils/generic_utils":130,"@tensorflow/tfjs-core":8}],128:[function(e,t,n){"use strict";function r(e,t){if(e.shape.toString()!==t.shape.toString())throw new Error("Shape mismatch: "+JSON.stringify(e.shape)+" vs. "+JSON.stringify(t.shape))}function i(e){return e instanceof p?e.value():e}var a=this&&this.__decorate||function(e,t,n,r){var i,a=arguments.length,o=a<3?t:null===r?r=Object.getOwnPropertyDescriptor(t,n):r;if("object"==typeof Reflect&&"function"==typeof Reflect.decorate)o=Reflect.decorate(e,t,n,r);else for(var s=e.length-1;s>=0;s--)(i=e[s])&&(o=(a<3?i(o):a>3?i(t,n,o):i(t,n))||o);return a>3&&o&&Object.defineProperty(t,n,o),o};Object.defineProperty(n,"__esModule",{value:!0});var o,s=e("@tensorflow/tfjs-core"),u=e("./common");!function(e){e.float32="float32",e.int32="int32",e.bool="bool"}(o=n.DType||(n.DType={}));var l=0,c=function(){function e(e,t,n,r,i,a,o){this.dtype=e,this.shape=t,this.sourceLayer=n,this.inputs=r,this.callArgs=i,this.outputTensorIndex=o,this.id=l++,null!=a&&(this.originalName=u.getScopedTensorName(a),this.name=u.getUniqueTensorName(this.originalName))}return e=a([s.doc({heading:"Models",subheading:"Classes"})],e)}();n.SymbolicTensor=c;var p=function(){function e(e,t){this.dtype=o.float32,this.shape=e.shape,this.val=e,this.id=l++,null!=t&&(this.originalName=u.getScopedTensorName(t),this.name=u.getUniqueTensorName(this.originalName))}return e.prototype.value=function(){return this.val},e}();n.ConcreteTensor=p;var h="Variable",d=function(){function e(e,t,n,r,a){void 0===t&&(t=o.float32),void 0===n&&(n=h),void 0===r&&(r=!0),void 0===a&&(a=null),this.dtype=null==t?o.float32:t,this.shape=e.shape,this.id=l++,n=null==n?h:n,this.originalName=u.getScopedTensorName(n),this.name=u.getUniqueTensorName(this.originalName),this.trainable=r,this.constraint=a,this.val=s.variable(i(e),this.trainable,this.name,this.dtype)}return e.prototype.read=function(){return this.val},e.prototype.write=function(e){return r(this.val,e),this.val.assign(i(e)),null!=this.constraint&&this.val.assign(this.constraint.apply(this.val)),this},e}();n.LayerVariable=d;var f=function(){return function(){}}();n.Serializable=f},{"./common":102,"@tensorflow/tfjs-core":8}],129:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});var r=e("../errors"),i=e("./generic_utils"),a=e("./math_utils");n.normalizeArray=function(e,t,n){if("number"==typeof e)return i.pyListRepeat(e,t);if(e.length!==t)throw new r.ValueError("The "+n+" argument must be a tuple of "+t+" integers. Received: "+e.length+" elements.");for(var o=0;o<t;++o){var s=e[o];if(!a.isInteger(s))throw new r.ValueError("The "+n+" argument must be a tuple of "+t+" integers. Received: "+JSON.stringify(e)+" including a non-integer number "+s)}return e},n.convOutputLength=function(e,t,n,r,i){if(void 0===i&&(i=1),null==e)return e;var a,o=t+(t-1)*(i-1);return a="same"===n?e:e-o+1,Math.floor((a+r-1)/r)},n.deconvLength=function(e,t,n,i){if(null==e)return null;if("valid"===i)e=e*t+a.max([n-t,0]);else{if("same"!==i)throw new r.ValueError("Unsupport padding mode: "+i+".");e*=t}return e}},{"../errors":107,"./generic_utils":130,"./math_utils":131}],130:[function(e,t,n){"use strict";function r(e){return Array.isArray(e)?e:[e]}function i(e,t){return e<t?-1:e>t?1:0}var a=this&&this.__assign||Object.assign||function(e){for(var t,n=1,r=arguments.length;n<r;n++){t=arguments[n];for(var i in t)Object.prototype.hasOwnProperty.call(t,i)&&(e[i]=t[i])}return e};Object.defineProperty(n,"__esModule",{value:!0});var o=e("../errors"),s=e("../types");n.pyListRepeat=function(e,t){if(Array.isArray(e)){for(var n=[],r=0;r<t;r++)n=n.concat(e);return n}return(n=new Array(t)).fill(e),n},n.pyGetAttr=function(e,t,n){if(t in e)return e[t];if(void 0===n)throw new o.AttributeError("pyGetAttr: Attempting to get attribute "+t+"with no default value defined");return n},n.pyNormalizeArrayIndex=function(e,t){if(null==e||null==t)throw new o.ValueError("Must provide a valid array and index for pyNormalizeArrayIndex(). Got array "+e+" and index "+t+".");var n="Index "+t+" out of range for array of length "+e.length;if(t<0){if(t<-e.length)throw new o.IndexError(n);return e.length+t}if(t>=e.length)throw new o.IndexError(n);return t},n.assert=function(e,t){if(!e)throw new o.AssertionError(t)},n.count=function(e,t){for(var n=0,r=0,i=e;r<i.length;r++)i[r]===t&&n++;return n};var u=function(){function e(){this.pythonClassNameMap={}}return e.getMap=function(){return null==e.instance&&(e.instance=new e),e.instance},e.register=function(e,t){this.getMap().pythonClassNameMap[e]=[t,t.fromConfig]},e}();n.ClassNameMap=u;var l=function(){function e(){this.enumRegistry={}}return e.getMap=function(){return null==e.instance&&(e.instance=new e),e.instance},e.register=function(t,n){if(e.contains(t))throw new o.ValueError("Attempting to register a repeated enum: "+t);this.getMap().enumRegistry[t]=n},e.contains=function(e){return e in this.getMap().enumRegistry},e.lookup=function(e,t){return this.getMap().enumRegistry[e][t]},e.reverseLookup=function(e,t){var n=this.getMap().enumRegistry[e];for(var r in n)if(n[r]===t)return r;throw new o.ValueError("Could not find serialization string for "+t)},e}();n.SerializableEnumRegistry=l,n.singletonOrArray=function(e){return 1===e.length?e[0]:e},n.toList=r,n.objectListUid=function(e){for(var t="",n=0,i=r(e);n<i.length;n++){var a=i[n];if(null==a.id)throw new o.ValueError("Object "+a+" passed to objectListUid without an id");""!==t&&(t+=", "),t+=Math.abs(a.id)}return t},n.isArrayOfShapes=function(e){return Array.isArray(e)&&Array.isArray(e[0])},n.normalizeShapeList=function(e){return 0===e.length?[]:Array.isArray(e[0])?e:[e]},n.toSnakeCase=function(e){var t=e.replace(/(.)([A-Z][a-z0-9]+)/g,"$1_$2").replace(/([a-z])([A-Z])/g,"$1_$2").toLowerCase();return"_"!==t[0]?t:"private"+t},n.toCamelCase=function(e){return e.length<=1?e:-1===e.indexOf("_")?e:e.replace(/[_]+(\w|$)/g,function(e,t){return t.toUpperCase()})};var c={};n.serializeKerasObject=function(e){return null===e||void 0===e?null:{className:e.getClassName(),config:e.getConfig()}},n.deserializeKerasObject=function(e,t,n,r){if(void 0===t&&(t={}),void 0===n&&(n={}),void 0===r&&(r="object"),"string"==typeof e){var i=e,s=void 0;if(i in n)s=n[i];else if(i in c)s=c[i];else if(null==(s=t[i]))throw new o.ValueError("Unknown "+r+": "+e);return s}var u=e;if(null==u.className||null==u.config)throw new o.ValueError(r+": Improper config format: "+JSON.stringify(u)+".\n'className' and 'config' must set.");var l=u.className,p=void 0,h=void 0;if(l in n?(p=(T=n.get(l))[0],h=T[1]):l in c?(p=(S=c.className)[0],h=S[1]):l in t&&(p=(C=t[l])[0],h=C[1]),null==p)throw new o.ValueError("Unknown "+r+": "+l);if(null!=h){for(var d={},f=0,g=Object.keys(c);f<g.length;f++)d[E=g[f]]=c[E];for(var m=0,v=Object.keys(n);m<v.length;m++)d[E=v[m]]=n[E];u.config.customObjects=d;for(var y=a({},c),b=0,w=Object.keys(n);b<w.length;b++)E=w[b],c[E]=n[E];return _=h(p,u.config),c=a({},y),_}for(var y=a({},c),x=0,A=Object.keys(n);x<A.length;x++){var E=A[x];c[E]=n[E]}var _=new p(u.config);return c=a({},y),_;var T,S,C},n.getExactlyOneTensor=function(e){var t;if(Array.isArray(e)){if(1!==e.length)throw new o.ValueError("Expected Tensor length to be 1; got "+e.length);t=e[0]}else t=e;return t},n.getExactlyOneShape=function(e){if(Array.isArray(e)&&Array.isArray(e[0])){if(1===e.length)return(e=e)[0];throw new o.ValueError("Expected exactly 1 Shape; got "+e.length)}return e},n.numberCompare=i,n.reverseNumberCompare=function(e,t){return-1*i(e,t)},n.stringToDType=function(e){switch(e){case"float32":return s.DType.float32;default:throw new o.ValueError("Invalid dtype: "+e)}},n.stringsEqual=function(e,t){if(null==e||null==t)return e===t;if(e.length!==t.length)return!1;for(var n=0;n<e.length;++n)if(e[n]!==t[n])return!1;return!0},n.unique=function(e){if(null==e)return e;for(var t=[],n=0,r=e;n<r.length;n++){var i=r[n];-1===t.indexOf(i)&&t.push(i)}return t},n.isObjectEmpty=function(e){if(null==e)throw new o.ValueError("Invalid value in obj: "+JSON.stringify(e));for(var t in e)if(e.hasOwnProperty(t))return!1;return!0}},{"../errors":107,"../types":128}],131:[function(e,t,n){"use strict";function r(e){return e=Array.isArray(e)?new Float32Array(e):e,s.tensor1d(e)}function i(e){return o.sum(r(e)).dataSync()[0]}function a(e){return i(e)/e.length}Object.defineProperty(n,"__esModule",{value:!0});var o=e("@tensorflow/tfjs-core"),s=e("@tensorflow/tfjs-core"),u=e("../errors");n.isInteger=function(e){return e===parseInt(e.toString(),10)},n.arrayProd=function(e,t,n){null==t&&(t=0),null==n&&(n=e.length);for(var r=1,i=t;i<n;++i)r*=e[i];return r},n.min=function(e){return o.min(r(e)).dataSync()[0]},n.max=function(e){return o.max(r(e)).dataSync()[0]},n.sum=i,n.mean=a,n.variance=function(e){var t=o.sub(r(e),s.scalar(a(e)));return o.sum(o.mulStrict(t,t)).dataSync()[0]/e.length},n.range=function(e,t){if(t<e)throw new u.ValueError("end ("+t+") < begin ("+e+") is forbidden.");for(var n=[],r=e;r<t;++r)n.push(r);return n}},{"../errors":107,"@tensorflow/tfjs-core":8}],132:[function(e,t,n){"use strict";function r(e,t,n){return("inboundNodes"===e||"outputLayers"===e||"inputLayers"===e)&&0===t&&"string"==typeof n}function i(e,t){if(null===e)return null;if("string"==typeof e)return s.toCamelCase(e);if("number"==typeof e||"boolean"==typeof e)return e;if(e instanceof Array){for(var n=[],a=e.length,u=0;u<a;++u){var l=e[u];r(t,u,l)?n.push(l):n.push(i(l,t))}return n}for(var c={},p=0,h=Object.keys(e);p<h.length;p++){var d=h[p],f=e[d];if("name"===d&&"string"==typeof f)c[d]=f;else{var g=s.toCamelCase(d);if(!s.SerializableEnumRegistry.contains(d)||"string"!=typeof f&&null!=f)c[g]=i(f,g);else{var m=s.SerializableEnumRegistry.lookup(d,f);if(null==m)throw new o.ValueError("Unkown value "+f+" for "+d+" Enum");c[g]=m}}}return c}function a(e,t){if(null===e||void 0===e)return null;if("string"==typeof e)return s.toSnakeCase(e);if("number"==typeof e||"boolean"==typeof e)return e;if(e instanceof Array){for(var n=[],i=e.length,o=0;o<i;++o){var u=e[o];r(t,o,u)?n.push(u):n.push(a(u,t))}return n}for(var l={},c=0,p=Object.keys(e);c<p.length;c++){var h=p[c],d=e[h],f=s.toSnakeCase(h);if("name"!==h&&"className"!==h||"string"!=typeof d)if(!s.SerializableEnumRegistry.contains(f)||"string"!=typeof d&&null!=d)l[f]=a(d,h);else{var g=s.SerializableEnumRegistry.reverseLookup(f,d);l[f]=g}else l[f]=d}return l}Object.defineProperty(n,"__esModule",{value:!0});var o=e("../errors"),s=e("../utils/generic_utils");n.convertPythonicToTs=i,n.convertTsToPythonic=a},{"../errors":107,"../utils/generic_utils":130}],133:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});n.version="0.5.3"},{}],134:[function(e,t,n){},{}],135:[function(e,t,n){var r=e("./lib/alea"),i=e("./lib/xor128"),a=e("./lib/xorwow"),o=e("./lib/xorshift7"),s=e("./lib/xor4096"),u=e("./lib/tychei"),l=e("./seedrandom");l.alea=r,l.xor128=i,l.xorwow=a,l.xorshift7=o,l.xor4096=s,l.tychei=u,t.exports=l},{"./lib/alea":136,"./lib/tychei":137,"./lib/xor128":138,"./lib/xor4096":139,"./lib/xorshift7":140,"./lib/xorwow":141,"./seedrandom":142}],136:[function(e,t,n){!function(e,t,n){function r(e){var t=this,n=o();t.next=function(){var e=2091639*t.s0+2.3283064365386963e-10*t.c;return t.s0=t.s1,t.s1=t.s2,t.s2=e-(t.c=0|e)},t.c=1,t.s0=n(" "),t.s1=n(" "),t.s2=n(" "),t.s0-=n(e),t.s0<0&&(t.s0+=1),t.s1-=n(e),t.s1<0&&(t.s1+=1),t.s2-=n(e),t.s2<0&&(t.s2+=1),n=null}function i(e,t){return t.c=e.c,t.s0=e.s0,t.s1=e.s1,t.s2=e.s2,t}function a(e,t){var n=new r(e),a=t&&t.state,o=n.next;return o.int32=function(){return 4294967296*n.next()|0},o.double=function(){return o()+1.1102230246251565e-16*(2097152*o()|0)},o.quick=o,a&&("object"==typeof a&&i(a,n),o.state=function(){return i(n,{})}),o}function o(){var e=4022871197;return function(t){t=t.toString();for(var n=0;n<t.length;n++){var r=.02519603282416938*(e+=t.charCodeAt(n));r-=e=r>>>0,e=(r*=e)>>>0,e+=4294967296*(r-=e)}return 2.3283064365386963e-10*(e>>>0)}}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.alea=a}(0,"object"==typeof t&&t,!1)},{}],137:[function(e,t,n){!function(e,t,n){function r(e){var t=this,n="";t.next=function(){var e=t.b,n=t.c,r=t.d,i=t.a;return e=e<<25^e>>>7^n,n=n-r|0,r=r<<24^r>>>8^i,i=i-e|0,t.b=e=e<<20^e>>>12^n,t.c=n=n-r|0,t.d=r<<16^n>>>16^i,t.a=i-e|0},t.a=0,t.b=0,t.c=-1640531527,t.d=1367130551,e===Math.floor(e)?(t.a=e/4294967296|0,t.b=0|e):n+=e;for(var r=0;r<n.length+20;r++)t.b^=0|n.charCodeAt(r),t.next()}function i(e,t){return t.a=e.a,t.b=e.b,t.c=e.c,t.d=e.d,t}function a(e,t){var n=new r(e),a=t&&t.state,o=function(){return(n.next()>>>0)/4294967296};return o.double=function(){do{var e=((n.next()>>>11)+(n.next()>>>0)/4294967296)/(1<<21)}while(0===e);return e},o.int32=n.next,o.quick=o,a&&("object"==typeof a&&i(a,n),o.state=function(){return i(n,{})}),o}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.tychei=a}(0,"object"==typeof t&&t,!1)},{}],138:[function(e,t,n){!function(e,t,n){function r(e){var t=this,n="";t.x=0,t.y=0,t.z=0,t.w=0,t.next=function(){var e=t.x^t.x<<11;return t.x=t.y,t.y=t.z,t.z=t.w,t.w^=t.w>>>19^e^e>>>8},e===(0|e)?t.x=e:n+=e;for(var r=0;r<n.length+64;r++)t.x^=0|n.charCodeAt(r),t.next()}function i(e,t){return t.x=e.x,t.y=e.y,t.z=e.z,t.w=e.w,t}function a(e,t){var n=new r(e),a=t&&t.state,o=function(){return(n.next()>>>0)/4294967296};return o.double=function(){do{var e=((n.next()>>>11)+(n.next()>>>0)/4294967296)/(1<<21)}while(0===e);return e},o.int32=n.next,o.quick=o,a&&("object"==typeof a&&i(a,n),o.state=function(){return i(n,{})}),o}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.xor128=a}(0,"object"==typeof t&&t,!1)},{}],139:[function(e,t,n){!function(e,t,n){function r(e){var t=this;t.next=function(){var e,n,r=t.w,i=t.X,a=t.i;return t.w=r=r+1640531527|0,n=i[a+34&127],e=i[a=a+1&127],n^=n<<13,e^=e<<17,n^=n>>>15,e^=e>>>12,n=i[a]=n^e,t.i=a,n+(r^r>>>16)|0},function(e,t){var n,r,i,a,o,s=[],u=128;for(t===(0|t)?(r=t,t=null):(t+="\0",r=0,u=Math.max(u,t.length)),i=0,a=-32;a<u;++a)t&&(r^=t.charCodeAt((a+32)%t.length)),0===a&&(o=r),r^=r<<10,r^=r>>>15,r^=r<<4,r^=r>>>13,a>=0&&(o=o+1640531527|0,i=0==(n=s[127&a]^=r+o)?i+1:0);for(i>=128&&(s[127&(t&&t.length||0)]=-1),i=127,a=512;a>0;--a)r=s[i+34&127],n=s[i=i+1&127],r^=r<<13,n^=n<<17,r^=r>>>15,n^=n>>>12,s[i]=r^n;e.w=o,e.X=s,e.i=i}(t,e)}function i(e,t){return t.i=e.i,t.w=e.w,t.X=e.X.slice(),t}function a(e,t){null==e&&(e=+new Date);var n=new r(e),a=t&&t.state,o=function(){return(n.next()>>>0)/4294967296};return o.double=function(){do{var e=((n.next()>>>11)+(n.next()>>>0)/4294967296)/(1<<21)}while(0===e);return e},o.int32=n.next,o.quick=o,a&&(a.X&&i(a,n),o.state=function(){return i(n,{})}),o}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.xor4096=a}(0,"object"==typeof t&&t,!1)},{}],140:[function(e,t,n){!function(e,t,n){function r(e){var t=this;t.next=function(){var e,n,r=t.x,i=t.i;return e=r[i],e^=e>>>7,n=e^e<<24,e=r[i+1&7],n^=e^e>>>10,e=r[i+3&7],n^=e^e>>>3,e=r[i+4&7],n^=e^e<<7,e=r[i+7&7],e^=e<<13,n^=e^e<<9,r[i]=n,t.i=i+1&7,n},function(e,t){var n,r=[];if(t===(0|t))r[0]=t;else for(t=""+t,n=0;n<t.length;++n)r[7&n]=r[7&n]<<15^t.charCodeAt(n)+r[n+1&7]<<13;for(;r.length<8;)r.push(0);for(n=0;n<8&&0===r[n];++n);for(8==n?r[7]=-1:r[n],e.x=r,e.i=0,n=256;n>0;--n)e.next()}(t,e)}function i(e,t){return t.x=e.x.slice(),t.i=e.i,t}function a(e,t){null==e&&(e=+new Date);var n=new r(e),a=t&&t.state,o=function(){return(n.next()>>>0)/4294967296};return o.double=function(){do{var e=((n.next()>>>11)+(n.next()>>>0)/4294967296)/(1<<21)}while(0===e);return e},o.int32=n.next,o.quick=o,a&&(a.x&&i(a,n),o.state=function(){return i(n,{})}),o}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.xorshift7=a}(0,"object"==typeof t&&t,!1)},{}],141:[function(e,t,n){!function(e,t,n){function r(e){var t=this,n="";t.next=function(){var e=t.x^t.x>>>2;return t.x=t.y,t.y=t.z,t.z=t.w,t.w=t.v,(t.d=t.d+362437|0)+(t.v=t.v^t.v<<4^e^e<<1)|0},t.x=0,t.y=0,t.z=0,t.w=0,t.v=0,e===(0|e)?t.x=e:n+=e;for(var r=0;r<n.length+64;r++)t.x^=0|n.charCodeAt(r),r==n.length&&(t.d=t.x<<10^t.x>>>4),t.next()}function i(e,t){return t.x=e.x,t.y=e.y,t.z=e.z,t.w=e.w,t.v=e.v,t.d=e.d,t}function a(e,t){var n=new r(e),a=t&&t.state,o=function(){return(n.next()>>>0)/4294967296};return o.double=function(){do{var e=((n.next()>>>11)+(n.next()>>>0)/4294967296)/(1<<21)}while(0===e);return e},o.int32=n.next,o.quick=o,a&&("object"==typeof a&&i(a,n),o.state=function(){return i(n,{})}),o}t&&t.exports?t.exports=a:n&&n.amd?n(function(){return a}):this.xorwow=a}(0,"object"==typeof t&&t,!1)},{}],142:[function(e,t,n){!function(n,r){function i(e,t,i){var p=[],h=u(s((t=1==t?{entropy:!0}:t||{}).entropy?[e,c(n)]:null==e?l():e,3),p),b=new a(p),w=function(){for(var e=b.g(f),t=m,n=0;e<v;)e=(e+n)*d,t*=d,n=b.g(1);for(;e>=y;)e/=2,t/=2,n>>>=1;return(e+n)/t};return w.int32=function(){return 0|b.g(4)},w.quick=function(){return b.g(4)/4294967296},w.double=w,u(c(b.S),n),(t.pass||i||function(e,t,n,i){return i&&(i.S&&o(i,b),e.state=function(){return o(b,{})}),n?(r[g]=e,t):e})(w,h,"global"in t?t.global:this==r,t.state)}function a(e){var t,n=e.length,r=this,i=0,a=r.i=r.j=0,o=r.S=[];for(n||(e=[n++]);i<d;)o[i]=i++;for(i=0;i<d;i++)o[i]=o[a=b&a+e[i%n]+(t=o[i])],o[a]=t;(r.g=function(e){for(var t,n=0,i=r.i,a=r.j,o=r.S;e--;)t=o[i=b&i+1],n=n*d+o[b&(o[i]=o[a=b&a+t])+(o[a]=t)];return r.i=i,r.j=a,n})(d)}function o(e,t){return t.i=e.i,t.j=e.j,t.S=e.S.slice(),t}function s(e,t){var n,r=[],i=typeof e;if(t&&"object"==i)for(n in e)try{r.push(s(e[n],t-1))}catch(e){}return r.length?r:"string"==i?e:e+"\0"}function u(e,t){for(var n,r=e+"",i=0;i<r.length;)t[b&i]=b&(n^=19*t[b&i])+r.charCodeAt(i++);return c(t)}function l(){try{var e;return p&&(e=p.randomBytes)?e=e(d):(e=new Uint8Array(d),(h.crypto||h.msCrypto).getRandomValues(e)),c(e)}catch(e){var t=h.navigator,r=t&&t.plugins;return[+new Date,h,r,h.screen,c(n)]}}function c(e){return String.fromCharCode.apply(0,e)}var p,h=this,d=256,f=6,g="random",m=r.pow(d,f),v=r.pow(2,52),y=2*v,b=d-1;if(r["seed"+g]=i,u(r.random(),n),"object"==typeof t&&t.exports){t.exports=i;try{p=e("crypto")}catch(e){}}}([],Math)},{crypto:134}],143:[function(e,t,n){"use strict";function r(e){for(var t in e)n.hasOwnProperty(t)||(n[t]=e[t])}Object.defineProperty(n,"__esModule",{value:!0}),r(e("@tensorflow/tfjs-core")),r(e("@tensorflow/tfjs-layers"));var i=e("@tensorflow/tfjs-core"),a=e("@tensorflow/tfjs-layers"),o=e("./version");n.version={"tfjs-core":i.version_core,"tfjs-layers":a.version_layers,tfjs:o.version}},{"./version":144,"@tensorflow/tfjs-core":8,"@tensorflow/tfjs-layers":109}],144:[function(e,t,n){"use strict";Object.defineProperty(n,"__esModule",{value:!0});n.version="0.10.3"},{}]},{},[143])(143)});
\ No newline at end of file
diff --git a/worker-installer.js b/worker-installer.js
new file mode 100644
index 0000000..373987c
--- /dev/null
+++ b/worker-installer.js
@@ -0,0 +1,9 @@
+if ('serviceWorker' in navigator) {
+  navigator.serviceWorker.register('worker.js', {
+    scope: '.' // IMPORTANT! Scope must be set on non-root apps for this to work
+  }).then(function(res) {
+    console.log('Service Worker successfully installed!');
+  }).catch(function(error) {
+    console.error('Failed to register Service Worker', error);
+  });
+}
diff --git a/worker.js b/worker.js
new file mode 100644
index 0000000..5e13786
--- /dev/null
+++ b/worker.js
@@ -0,0 +1,36 @@
+var TAG = 'not-hotdog';
+var RESOURCES = [
+  './',
+  'manifest.json',
+  'roboto-black.ttf',
+  'app.css',
+  'images/hotdog.png',
+  'images/cross.png',
+  'notification.ogg',
+  'worker-installer.js',
+  'tensorflow.js',
+  'app.js',
+  'model/mobilenet.json',
+  'model/mobilenet.classes.json'
+];
+for (var i=1; i<=55; i++) RESOURCES.push(`model/group${i}-shard1of1`);
+
+// Let's start by caching all resources on install
+self.addEventListener('install', function(event) {
+  console.log(`Service Worker installed for ${TAG}`);
+  event.waitUntil(function() {
+    return caches.open(TAG).then(function(cache) {
+      return cache.addAll(RESOURCES);
+    });
+  });
+});
+
+// When devices asks for a resource, we fetch it from cache (if available)
+self.addEventListener('fetch', function(event) {
+  event.respondWith(
+    caches.match(event.request).then(function(res) {
+      // Return cache, else fetch it from the network
+      return res || fetch(event.request);
+    })
+  );
+});