Skip to content

Implementations of the fastml-science bechmark models, including a standard Keras (float) and QKeras (quantized) implementations.

License

Notifications You must be signed in to change notification settings

fastmachinelearning/fastml-science

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

38 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fast Machine Learning Science Benchmarks

DOI Code style: black

Implementations of the fastml-science benchmark models, including a standard Keras (float) and QKeras (quantized) implementations.

jet-classify

Requirements:

Python 3.8

conda env create -f environment.yml

Training:

python3 train.py -c <config.yml>

Upon training completion, graphs for the ROC for each tagger, are saved to the output directory, along with a .h5 saved model file.

The benchmark includes a float/unquantized 3 layer model as well as a uniformally quantized 6b model

Sample Runs

Training Float Baseline:

python3 train.py -c float_baseline.yml

Alt text

Model test accuracy = 0.766

Model test weighted average AUC = 0.943

Training Quantized Baseline:

python3 train.py -c quantized_baseline.yml

Alt text

Model test accuracy = 0.764

Model test weighted average AUC = 0.941

beam-control

WIP

sensor-data-compression

WIP

About

Implementations of the fastml-science bechmark models, including a standard Keras (float) and QKeras (quantized) implementations.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published