diff --git a/committee-steering/meetings/007-2023-07-27.md b/committee-steering/meetings/007-2023-07-27.md new file mode 100644 index 00000000..1a78478b --- /dev/null +++ b/committee-steering/meetings/007-2023-07-27.md @@ -0,0 +1,399 @@ + + +# Akash Network - Steering Committee - Meeting #7 + +## Agenda + +- Review / discuss / address any community feedback, grievances, or conflicts. +- Review [projects list](../projects-list/README.md), prioritize / add / remove if necessary. +- A look at open discussions in [Github Discussions](https://github.com/orgs/akash-network/discussions). +- Open up to any other questions from the community. + +## Meeting Details + +- Date: Thursday, July 27, 2023 +- Time: 11:00 AM PT (Pacific Time) +- [Recording](https://2uw2d6ylfq3v2btiav5izgx4kvsgcc2s72bb5vvymb5byck4jsca.arweave.net/1S2h-wssN10GaAV6jJr8VWRhC1L-gh7WuGB6HAlcTIQ) +- [Transcript](#transcript) + +## Committee + +- Adam Bozanich, Overclock Labs +- Anil Murty, Overclock Labs +- Artur Troian, Overclock Labs +- Greg Osuri, Overclock Labs +- Scott Caruthers, Overclock Labs + +## Participants + +- Addo Smajic +- Andrew Gnatyuk +- Anil Murty +- Ariel Leza +- Artur Troian +- Boz Menzalji +- Damir Simpovic +- Denis Lelic +- Deval Patel +- George Pro +- Greg Osuri +- Javier Suarez +- Jeffrey Lipps +- Jigar Patel +- Joao Luna +- Julius Haus +- Keenan Nemetz +- Lowell Tarek Abbott Vidal +- Max +- oo o +- Robert Del Rey +- Rodrigo Rochin +- Scott Hewitson +- Tyler Wright +- Zach Horn + + +## Notes: + +- Anil reviewed the structure of the meeting for any new members. +- Anil started the meeting by asking if there any feedback on how things are going in the community or any grievances? +- There was no feedback from the community. +- Anil talked about the Projects list. +- Greg talked about Mainnet 6. This is a huge update for Akash. +- Anil mentioned that ZJ was able to deploy SDXL. Shoutout ZJ! +- Greg reiterated this fact, and made it clear that the Passage investment was from Overclock Labs. +- Passage will be a big user for Akash, and they are doing a lot of work. +- Sigs and Wgs coming up next week include sig-economics, sig-support, sig-chain, and others. Stay tuned in discord to stay up to date. + + +### Projects: + +- GPU Incentivized Testnet ended a week ago. +- There are a few technical issues that are covered in sig-support. The issues are 104 and 106. +- After testing out these issues, there will be a testnet for take pay and stable payments. These things will be figured out in the next few weeks. +- Max aske if take pay will be before or after the Mainnet. All the changes to GPUs and take pay will happen before the Mainnet. +- Mainnet Upgrade is tracking for the first or second week of August, depending on testing. +- Anil talked about the hackathon and the events which are getting a lot of work right now. +- Luna spoke about EthParis. +- Anil mentioned that they are talking to a team working on decentralized internet project that wants to be apart of the Akash Hackathon. +- Anil talked about documentation site. Luna and his team are working on docs 2.0. +- Scott from Overclock Labs has been working on developer documentation. +- Anil talked about analytics. Overclock is talking with Cloudmos about analytics. Zack is working on analytics for the entire Cosmos, including Akash. +- Max and team are working on a schema for provider attributes. Anil has a spec that he is working on. Feel free to give feedback on this. +- Max made a PR for the schema today. Here is the [PR](https://github.com/maxmaxlabs/cloudmos-config/pull/1). The main goal is to have a base schema that can evolve over time. +- Anil talked about AuthZ. There are two sides of this. Cloudmos team is going to work on AuthZ after provider attributes. +- Anil mentioned that Overclock Labs and Cloudmos are talking about sharing code. This code would be in AkashJs. The conversations are early. +- Anil talked about content moderation. This was on pause from the Praetor team, while they focused on GPU Testnet. +- Jigar mentioned that after providers are stable on Mainnet +- Most of the work for GO Client Library is pending to wait for discussion, and waiting for the Mainnet to happen. +- Artur mentioned that there is a delay on his side to review code. There are some things that need to be addressed before moving on. +- Anil reiterated that GPUs and Mainnet take priority right now. +- Anil wanted to plant a seed about a few things. + - There has been discussion of trying to figure out a SPOT marketplace for GPUs and CPUs. Many of the workloads are batchy in nature, so resources are utilized for only part of the day. The thesis is that companies can bring supply onto the market while they are not using it. In order for this to work, there needs to be a few changes. Need to make sure that inventory removable at certains times does NOT affect uptime. Also need to figure out how this will all work from a logistical standpoint to user experience optimization. + - As part of the GPU AI Incentivized Testnet, it has been important to find supply and demand. There are some tenants that have been identified that will be paired with providers. There is a thought to start a project to test AI foundational models. + - Greg mentioned that there is a notion that decentralized networks cannot Train a model. If we can prove that wrong, that would be great for web 3 in general. The whole process will be documented to share the story with others. It will be incredibly helpful for Akash, and will help build tools for improving the experience. + +### Github Discussions: + +- Anil talked about discussions around analytics. +- Boz mentioned that John has to update this proposal based upon Akash's needs. +- Cheng put in a proposal for inflation pool, and this has since passed on chain [here https://www.mintscan.io/akash/proposals/211. +- Boz and Anil will look at the discussion from Weco [here](https://github.com/orgs/akash-network/discussions/276). +- Boz will respond to the person about building a Delegation platform in the discussion here: https://github.com/orgs/akash-network/discussions/280. +- Many people mentioned that mintscan version 2.0 is very nice. + + +**Open Discussion:** + +- Andrew asked about Passage investment announcement. This was funded by Overclock Labs; not the Akash Network. + +### Action Items: + +- Luna will turn the docs site back on for Greg to give feedback. +- Artur to review code for GO Client Libraries. +- Anil will pusblish a discussion with next steps after meeting with the team for training an AI model on Akash. + + + +# **Transcript** + +_This editable transcript was computer generated and might contain errors. People can also change the text after it was created._ + +Tyler Wright: Welcome everybody to steering Committee Monthly meeting Number seven. The steering committee is a culmination of all the special interest groups and working group meetings that happen throughout the month. Where again the steering committee comes together to go over projects that are actively being worked on talk about special interest groups and working groups, talk about any new or relevant discussions inside the github discussions. And then again, open up for any other conversation, from members of the community. so again what usually happens, the Neil walks us through the agenda and I'll be taking notes in the background, so when they'll take it away, + +Anil Murty: Thanks Good morning afternoon evening. I'm gonna go into my screen so we can start looking at the agenda and things that you want to discuss. Should be coming up now. + +Anil Murty: Alright, so for those that are joining us for the first of all, welcome to the community. The way we do things typically is like we have a bunch of projects and working groups that work asynchronously and then meet as needed. And then on an approximately monthly basis, we review pretty much all the projects and then make decisions about, governance of the overall open source community, as well as decisions about priorities of projects and things like that. So typically what we do on these meetings is we sort of go for agenda items. We review and discuss any community feedback in terms of how things are working within the working groups, as well as sigs or any other feedback really in terms of how we can improve the process going forward. + +Anil Murty: We reviewed the list of projects that we have prioritized as a group and see if there are things that need to be shuffled around or if the status of some of the projects needs to be changed. We review any github discussions and see if there's things we need to do to move things forward or create any sort of work groups out of them. And then we open it up for any other open questions from anybody? That's on the call that wants to bring up something that hasn't been discussed. outside of these four things, I have a couple of things that I want to talk about as well. These are sort of new projects, it'll probably fit in the project list area, so I'll bring them up in that point. But that's kind of the agenda before we get into that. Is there any sort of general questions about it? + +Anil Murty: Cool and take that. As I know, does anybody have any sort of feedback or grievances or conflicts? They want to bring up about the way things are working in the communities as such Any sort of? Issues that you see with the way things are being done. + +Anil Murty: yeah. We typically get pretty much no feedback on this one. So it's really a good thing. Hopefully people aren't holding back. So yeah, it's good to know that overall things seem to be going and people are happy with the way things are working. So we're going to move on to the next item which is reviewing the list of projects that are currently in the backlog here. So for people that want to look at this list it's that URL there. You can also get to it from the notes within the steering committee. Read me as well. Here's a list of projects and it typically goes from top to bottom in this table here. And for each one, we have, a short description, the committee group that's working on it, as well as the current state of the project. So the very first one here is GPU support obviously that is front incentive for us. Yeah, just realized I may not be sharing. My entire screen showing up. Should. + +Anil Murty: Yeah. Looks like I was just sharing the tab and not the window. So now it should be Visible. cool. So this is the list of projects and you can get to this again from the steering committee notes. The very first project here is GPU support. Most people on this call should be familiar that we've been very focused and hands down on That incentivize testnet and ended about the books approximately when one week ago. and it's been overall, very successful, so huge. Thanks to everyone, obviously the core team Artur with his and everybody from the 14 that's supporting it, as well as + +Anil Murty: Thanks to the community that has really helped us make the successful in terms of Getting people to participate, reviewing all the submissions and then answering any questions on discord Channel. So huge, thanks to everyone there in terms of the current status and made things go from here, one of the goals. So obviously there was three or four goals, we had established for the GPU Incentivized tested, one of it was to test the product changes and the code changes itself. the second was to, build out more AI models within awesome Akash. + +00:05:00 + +Anil Murty: And then the third was to generally create content to evangelize Akash and position August akash as a valid provider for AI models. And I think we've done very good on all three of those fronts specifically talking about the technical portion of this, which is identifying any issues ahead of There were two main issues that came up sort of in the context of testing, Both of those issues are documented in our support repo and so, the most critical of these issues is issue number one or four out here. Where, a certain type of SDL was causing the provider to restart. And so, I believe this is closed now and there's a fix in place. We are currently testing this with + +Anil Murty: a sandbox environment with, first as initial provider that the Overclock core team is testing and then we'll be endless the help of a few of the insiders for testing this out as well. So that is currently in progress. There is another open issue. I believe that's open issue. 106, that is also getting a code fixed from Arthur. And so these two are kind of the main two issues that are outstanding in terms of feedback or issues that came out of the testnet and the incentivize just need as well. So I hope is that we'll be able to test these out this week and verify that they're fixed and then have them ready for the minute upgrade. I don't know, Artur is anything else you would like to add in terms of technical feedback that you've gotten from the inside device chestnut? + +Artur Troian: another max actually, we cover, all of the things related to issues and features in the support discussions + +Anil Murty: Thank So Basically the plan right now is we're going to run this sandbox test with these that verify these few bug fixes and code changes. And then, the next thing we're going to work on is a smaller test net, an upgrade to a test mid network, where we'll test out stable payments as well as the economics 2.0, which is support for take pay. And so those changes are something that we are targeting before the minute upgrade. So test those out in a small testnet and then have them verified and ready for inclusion in the next minute of day. And so those kind of are reflected in these next two items which was an implementation right now, whereas probably something that'll change to beta once we have this testnet up and running with the changes for it. So, that's those kind of the top three items out here, any questions on those ones before you move down the list here. + +Anil Murty: Go at Max. + +Max: Level payments. + +Anil Murty: Yeah, great question. Answer is no. So the plan is basically, we're going to first verify the fixes for these GPU issues, which I think is ongoing right now. Next, we're going to take the same sandbox environment or, the testnet environment depending on what the deputy team decides on and updated with the code changes for stable payments as well as make and then we're going to run a smaller testnet with that and this is going to all happen before And then the intention is that when we do the main IT upgrade which right now is potentially targeted around the 7th of August but that could change as we find out things in the next few days. So when we do that meeting updated basically include all of these changes so support for GPUs as well as make and take pay and stable payments. + +Anil Murty: Cool. + +Anil Murty: Any other questions? + +Anil Murty: Yeah, so as I mentioned right now, we're tracking a mainnet upgrade that includes all of these three somewhere around the second week of August. But we'll keep you all posted to discard as well as, any other meetings on how that is progressing and if you know that moves a little bit later, Moving on to the next set of things and I kind of grouped these based I shuffled the table around just a little bit this morning based on how things have been progressing in terms of these projects. And I feel like the hack account is something that is been doing a lot of work in terms of getting ready for being able to do this in August. And I know there's been a lot of discussions around events. I know there was participation from community folks in a couple of events recently as well. So, I don if you want to call through your + +00:10:00 + +Anil Murty: Doesn't look like Adam watching this because everybody else it wants to talk about the events or HACKATHON or certification program stuff at all. + +Joao Luna: Is there anything specifically to be of events that you want to hear about? + +Anil Murty: I just generally like if anybody wants to share feedback on the events that the community folks participated in that will be great to hear. And then on the hackathons, specifically, if anybody wants to share any updates in terms of, we're looking how we're tracking in terms of the timeline as well as the identifying, the projects that we're going to open up as part of that. + +Joao Luna: I think I can speak a little bit on the events at least from the parents portion of it. I went pretty well a bunch of cool stuff. Spanned from there. + +Joao Luna: I think that I'm aware specifically, one of the things that I participated in the acathon and sorry, my voice is kind of Bad. Because I catch something at the conference But So I did participate in the HACKATHON, the ink global echathon and what we built is something that maybe we might be bringing to the Akash network if it makes sense. + +Joao Luna: it was basically a tldr it's a backup tool for a bunch of database management systems that can run as a sidecar container. That is something that we could As part of the Akash network provider offering to allow them to data through two IPFS file code. So the idea is to offer kind of a decentralized storage backup solution. It provides us once. Yeah, I can give more details on that on discord. That's one other things also made a lot of guys. + +Joao Luna: On the osmocon and the cosmos events. It was pretty cool. I gave a talk about the cash network. It was also very very interesting. + +Joao Luna: A lot of people had questions wanted to see Demos and also the opportunity at Global to devote the product to a bunch of people. I left a message on the discourse saying If anyone is interested in deploying their projects on the decentralized cloud, they could DM me and I got a bunch of responses and it allowed me to go through the venue and show people like the cloud most tool, the cash console, and just spread the world a little bit, that was cool. The most security thing at osmocon the presentation was pretty awesome. Something to look forward to the mesh security, take notes and go research a bit about it. It's very, very interesting. What else? + +Joao Luna: And there's a bunch of other stuff. I'm just not really remembering. Also met a guy called The Cash since it's a somewhat popular Indian name. It was fun. + +Joao Luna: And yeah, pretty much is also met someone that he said, He used to maintain the omnibus repo back in the day. + +Joao Luna: I think those are kind of the highlights. + +Anil Murty: Thanks Yeah, that's a really good update. So some seems like overall Things prevent pretty well Eat conference, so that's great. Awesome. Yeah if Adam wasn't on the call he'd probably give us a more detailed update in terms of the next coming up conference is but I don't know If you have any other updates on that or the icon, go ahead. I know you already hand up video. + +00:15:00 + +Tyler Wright: I think Luna gave a great update. I know against some other folks went to Adam Berlin and folks in the insiders and had a great time there was an insider office hour to just wrapped up a little bit ago. And I know that Adam mentioned on that call that the HACKATHON registration should be starting next week. So folks should look out for that. And again, now spread the word if they can. And then there should be a working group HACKATHON session that he's looking to plan either tomorrow or again, early next week for folks. So, again, look out for the events and discord, if you want to get involved and the next steps for the HACKATHON, + +Anil Murty: Pretty Cool and The list of projects for the HACKATHON is still not finalized, right? That's something that's going to be worked on in the next few days. + +Tyler Wright: I believe so yeah I think they have five general tracks but I think they're gonna continue to fine, tune it over the next couple of days for sure. + +Anil Murty: Okay, great. Yeah, we've been talking to one with three project that is doing work in terms of basically getting network coverage to essentially a decentralized, Internet coverage kind of project and they were interested in partnering with us on the sacathon to see if basically have your four different types of nodes and they wanted to see if As part of the HACKATHON, we can make one of the tasks, spinning up one of the nodes or these for their network onto So something that we're talking to them about we'll try to write up a document or something that can be cured for consideration as part of the hackathons. Go ahead, Luna. + +Joao Luna: I just remembered something as well. I think they are from the company that's building open registry or at least know And they told me that they are making some progress. So just kind of an update on the open registry thing. Not sure if it traveled relevant or not. But yeah, I think it's at that they are working on it. It seems + +Anil Murty: Was it one of those folks that highlighted there? + +Joao Luna: From the name, second. And not sure. + +Anil Murty: Yeah, because they actually did a presentation in the last six clients meeting on their latest this stuff,… + +Joao Luna: Yeah, I remember yeah. + +Anil Murty: so it's probably one of them. I think there's just two people that work on it right now, so that's cool that you made them differently. If you weren't on it, definitely check out the recording from this meeting because they did a cool demo of their latest client and UI changes and stuff. Cool. + +Joao Luna: But it was one month ago, right, that they did it or was it recently? + +Anil Murty: I don't. Yeah, I think it was probably three weeks ago. I want to say Maybe. + +Anil Murty: It's Okay, so it looks like HACKATHON is going fine in the entirely. Even did anybody have anything to share on certificate program about certification program? Probably not. You can hold off for next time. + +Anil Murty: let's see. Dockside migration I know there's a bunch of efforts going on this one. So Luna and Team obviously are working on the new layout for the new DOC site and have a staged environment that people can check out and see kind of what the workflow looks Separately, adding a lot of Scott on our side has been adding a lot of documentation internally for that's more developer focused. So I think the eventual goal here is to sort of merge these two efforts together so the content from our team going to Luna's team to get incorporated into the New York site. And then eventually get the dockside, updated on the current website, and replace it essentially with Docs.com overall plan still, or any other updates to add from your side, Luna title. + +Tyler Wright: I think that's very accurate up there. I think we're going to again, push it eventually to be a network slash docs, but the next phase is, there's going to be a working group session that's going to be set up for 9, am on Monday, where Luna Scott and some other members of the community are all going to come together to talk about strategy. I know again, Scott and other members of the community, contribute to developing new documentation, and we have to figure out a documentation freeze. So that Luna and his team can do it last, migrations and push us into Hugo and then we can figure out a strategy for the transfer. So again, there'll be a discussion about that on coming Monday. And it's already on the calendar, but I'll be a bad added to the discord events today. + +00:20:00 + +Anil Murty: To Thank you Share that update. So great looks like by the next year and company meeting, maybe we'll have a new dockside. We'll see. + +Anil Murty: The next one here is on chain analytics so there's everything related to seek analytics kind of There's been a couple of updates here. there was almost in team have done an amazing job of supporting the GPU testnet as well as repeat or team on the provider side. thank you both teams for doing the work to support the testnet. It's been amazing for us to be able to utilize that data in terms of social sharing that And then I think the current plan is we're. So the one person that supported that submitted a proposal Zach, I don't know, Zach's on the call, let's look at the list of people here. Doesn't look like he's on the call right now, but Zach with the key. + +Anil Murty: He is working on an indexer and he's kind of building a solution makes a solution for the entire cosmos ecosystem so it's not specifically Akash but he's going to work on one that is going to support our cash as well. And in separately we are talking with the cloud most team to see, how we can work with them utilize their indexer as well. So that's kind of what's going on going Greg. + +Greg Osuri: What are the timelines in the documentation? I + +Anil Murty: The dockside, migration. yeah,… + +Greg Osuri: Yes. + +Anil Murty: so I think the Tldr and that is basically + +Anil Murty: Next week is a meeting where. Luna and Scott and others are going to meet and talk about how all the content that Scott has been developing. That is open source developer focused can get merged into the new site that Luna 19 have developed and so the hope is that once again out that We should be able to get the new dockside updated to the website in the next few weeks. Not at least might take on that. + +Greg Osuri: And what happened to the staging server? it was up for a little bit but it's gone. You want to bring that back so I can start giving early feedback. + +Anil Murty: Good question. Luna, do you know + +Joao Luna: Is it all time you haven't checked? It was a development environment but I can bring it up for feedback if I turned around it's probably to save some Pakistan. I can bring it up as again. No problem. + +Greg Osuri: Yeah, that'd be great. Thank you, appreciate it. + +Anil Murty: It's good, moving on to the next item. Here is provider attributes. So I move this up in the list of things a little bit because Max in the town was team stepped up inside, they are going to test out implementing, the rough spec that we have outlined a while back. The spec, we're document right now is shared on the WG Provider Attributes, discord channel, anybody that wants to look at it, go ahead and request access and identity to you. The only reason I'm keeping it closed right now is because it's not fleshed out completely and I also want to know who's reviewing it so I can get feedback from you. But I mean, while Max in team are + +Anil Murty: Working on implementing a schema, in cloud, most to support the spec in one. And one of the goals are there is to test, sort of how this would work. And if any changes are needed from the schema, in order to support more flexible, creating mechanism. But that's kind of what's going on here. That's why this is an implementation. So I don't know. Max, if there's anything else you want to add to that. + +Max: I created a pull request today. I put the file in one of our public repos to be somewhere else, just did it for simplicity. And I just shared the pull request with bunch of. Comments on why certain things are this way or the way that we propose to do it. So, If anyone wants to Comment on it or Can I guess contribute or discussed in that pull request? Even though It's merged in the master branch so yeah. The main goal is to have this. Base schema that we can start working on. And make it evolve with time. + +00:25:00 + +Max: based on what we learned while using it and So yeah, that's a Just look at it. Give us your comments and I'm already working on A form on cloud nose. So that providers will be able to edit these attributes based on a schema. And therefore, if ideally all the providers provide the same information in the same attributes, we should have a fairly standardized set of information for each of the providers. + +Anil Murty: Thank you, Max. Yeah, this super important because it'll definitely improve the user experience for clients deploying on right now, it's nearly impossible to Implement any serious filtering in terms of geographic locations, or anything? That's hard because the attribute keys are all over the place. So thank you. The next one here is authorized spend management. + +Anil Murty: This one used to be called Console or three before I changed it to just rgen in general, because we've been talking with the cloud most team on this a little bit and the spec, which is linked from here. If you look at it it calls out a bunch of features on two sides of this equation. So it calls out features to implement support for odds. Z for people that are essentially using Essentially, the grantee and then also the grantor. So people that are going to be giving out these funds and being able to look at what wallets have been granted in a single view, be able to see amount them on that is spent and remaining, so on and so forth, and be able to revoke and manage all the while it's from a single page. + +Anil Murty: So in speaking with a cloud nose team, one of the things we realized is that Cloud, most supports some of these things and doesn't support some of them. So, cloudmose is going to take a stab at this, once to get through the proper attribute stuff unless there's something else that comes up, So, that's kind of the plan or not Z for now. And one of the things I wanted to talk about today, which kind of I think fits really well with this is, we've also been talking with the Cloud most team about. Potential reuse of code between, Console and Carlos. So, talking about areas where + +Anil Murty: if there are certain things that are implemented by console today that are differently implemented from the akashias repo, getting those into Apacheous and similarly, our side as well. We'll give a more detailed update on this. It's still pretty early conversations that they're having, but there is essential idea, is to make Akash JS, which is our JavaScript library, a place where a lot of the common functions reside that any clients including almost and console can utilize for common functions. So, that's another update that I just want to share on this call. + +Anil Murty: The next one's content, moderation. I know this was kind of fun pause, because there was a So the primary people driving this after we defined the specs and the PRD for this collectively, as the primary team driving, this was the Plateau team and the prior team obviously has been very busy, the GPU testnet and modifying their onboarding client to support bringing you on GPUs and all that works. So this has been kind of fun. Pause, I think the plan was to resume this in the next week or two weeks. I don't expect a whole lot of updates right now. What, jigar they will. If there's anything else you want to add good. + +Anil Murty: Thank The next one's client libraries. I don't think there's been a huge amount of work on this recently. But at least from what I've seen. I know, Luna, you and your team were leading effort on the GLOW client library. is there anything you want? You would like to share. + +00:30:00 + +Joao Luna: Yeah, we are currently focused mainly on having it ready for and supported and support the Tesla. So most of the work is pending the release is spending because it's waiting for the discussion as well as the test that two main net. + +Anil Murty: good after + +Joao Luna: So our efforts are really on making sure that the client libraries are working. It does not. + +Artur Troian: Yeah, on the coin side, there's a little bit of July from my part to review code and do suggestions for that. I frankly saying I don't have an idea to have hopefully once I'm gonna loaded with the GPU and my net break To the final point, I will be able to take a look at that and give something, guys, because they're quite a few things that needs to be addressed before. + +Anil Murty: Thank you Yeah, that makes sense and I don't think there's a huge rush. Obviously, the main date upgrade and bug fixes and testnets take a priority over this. So make sense all right, I don't think this much point in discussing a lot of these other projects, unless anybody specifically wants to talk about one of these, Because I don't think there's any active work going on Anything else anybody wants to bring Out of this list. + +Anil Murty: Okay if there's nothing else you want to bring up, there's a couple of things I want to kind of plant a seed for these are not things that have a spec or anything right now, but it's something that we've been discussing a little bit within the community. + +Anil Murty: So one of these is basically, coming up with a specification for what needs to happen before Akash can support a spot marketplace for providers. So to provide some context, in this one of the thesis is we have on our side is basically in addition to, being a marketplace for people that have their own compute, whether it's GPUs that you own, we would also like Akash network to be the place that people can bring their cloud resources to. So, this is a situation where, a lot of people today especially in the case of GPUs, if you want to get access to the most latest and greatest GPUs like the A100s and the H100, it's currently most of the big cloud providers like Amazon, Google, and as your they're going to charge you. + +Anil Murty: Basically ask you to commit to a lengthy contract and typically these run anywhere from one year to three years. by doing so and committing to this what's called a committed use discount, you get a discount on the regular price that you would get if you were just paying on demand. So, this concept of reserved instances in AWS or committee use discounted GCP. This is what a lot of small to big companies end up committee to and it's a big price for them to pay. when you think about AI and management type of workload, specifically many of these tend to be batchy in nature and lots of times people that are running. Inference or training, their resources, are utilized, only for part of the day. + +Anil Murty: And so, one of theses we have is that if Akash had a way of letting these companies that have committed to these long-term contracts, bring their resources to Akash during the times, where they're not utilizing it for their own needs, then that could potentially be a good source of supply, especially for some of these hard to get GPUs. Right. So, this basically a situation where rather than somebody that is been a provider or a data center company operator for their full-time business coming out to Aka Now we're actually looking at people that are potentially tenants that have leased compute capacity from some other cloud. Provider them becoming providers and akash because they want to monetize or reduce the cost that they take on for the computer capacity that they've committed to + +Anil Murty: So this is the idea of its plot marketplaces and so one of the things we realized over there is in order for this to work, we have to do a couple of things. One is, we think on the provider side we should be okay because there is a way for the provider to take their workloads off the network in the times when they need to use them for their own needs. And, one of the things to verify there is, whether it affects the up time that we display for that provider. We don't think it does affected right now, what we need to verify that And then on the client side is the more tricky thing which is making it very clear to users that when they are selecting providers. They know that there are certain providers that will be more cost competitive or certain resources that'll be more cost competitive. + +00:35:00 + +Anil Murty: But these are interruptible resources so if you're running a workload on this lower Cost Spot type instance, there's a chance that your workloads could be interrupted without any notice and hopefully the providers are able to publish a schedule for when these resources are available and so figuring out how all this is going to work from both from our logistic standpoint, as a list, from implementation standpoint, and user experience standpoint, is something that I think we need to look at. And so I'm going to take us again for this, in the next, Three, four weeks, maybe and see if we can start a working group or something around this. + +Anil Murty: That's kind of one thing I wanted to bring up. I don't know if there's any thoughts or questions around that whole idea of spot providers in, if everybody has any thoughts around whether it's a good idea, bad idea or any thoughts that people have had. + +Anil Murty: I'll take that as a strong agreement. The other thing I wanted to bring up was as part of the Testnet for GPUs and AI models. One of our goals has been to identify obviously providers. That can, be good providers Bring to the mainnet and that's been an ongoing effort but then separate from it. We have also been closely trying to find folks that are building AI startups or working on AI projects. That could be good tenants for these providers when they're on the minute. And so a few conversations that we've identified some tenants and we're going to be paying up them up with providers. But then separate from that we've also come across a couple of folks that are working on training foundational models, as well as fine tuning existing foundational models. + +Anil Murty: So being able to take an existing model and apply something like loader techniques to it or training, a foundational model from the Using a different data set is something that It's attractive to us for two reasons. One is because if we can demonstrate that a foundational model can be trained on a cash network, then it becomes an example for other people that are wanting to do this or doing this on other clouds. And then the second reason is that we have a workload that is running for extended period of time. So in some cases where somebody wants to take a stable, diffusion base model is the one or two or something and then train it on a different data side. + +Anil Murty: This could, take hundreds or thousands of hours of GPU resources and that would be a very good demand working on the demand side. And then, if you took something even more complex than that, You could be looking at tens of thousands of hours of q100 utilized for it. So we're having conversations with a couple of folks on that side that can help us with this that are very experienced with training AI models. And so, the outcome of this will likely be a governance proposal that will be submitted on the discussions. They'll talk about what the benefits of this would be to our cash community as well as what the cost would be. So look for that as well. That is the two main other things I wanted to share. Is there anything else to bring up in the projects area before we kind of talk about discussions and governance proposals? And then wrapped things up, + +Greg Osuri: Yeah, it's gonna add a little more light to the foundational model. I think this is an incredible opportunity. There is a notion that decentralized networks are not good for training. And the notion that decentralized networks are only for inference. finally, So by showing that you can actually, train a foundational model, we break the narrative. And as huge not only for a cache but crypto in general, right? That. + +Greg Osuri: Content, we get out of it tutorial that or the learnings, we + +00:40:00 + +Greg Osuri: So yeah, that's and the marketing power we get with it, as we are as we can go with this A, Documents and ensure machine learning folks who can do it. So, it's overall. I think it's going to be incredibly helpful for Akash and also makes us build amazing tools to make this whole process easier. So, both not only from marketing and ordering from awareness, but also for my product standpoint things, very great. Overall. + +Anil Murty: Thanks Appreciate the additional context and super excited. So yeah, the next steps on that is I'm going to work closely With the team that is potentially going to work on this with us partner with us on this as building an AI app right now. And we'll publish a covenants, proposal that will include kind of a high level description of what the project is going to be and what the outcomes will So stay tuned for that. Anything else to bring up on Anybody last time? Going once, going twice. + +Anil Murty: I guess we'll jump back here and look at what's going on in Github discussions. In terms of any new governance proposals that are submitted. Out here. So, let's see. + +Anil Murty: I guess last time we were here, we talked until about here. Since then, there's been a proposal for bringing Metamask to Akash in the rest of Cosmos. I don't know, boss, your super close to this effort and you want to talk about this at all. If not, no worries. + +Boz Menzalji: Yeah. the action items are on John to get back to us with an updated proposal based on the requirements that we shared with them, this is going to change us a little bit. So I'll ping John to update this proposal based on our needs. + +Anil Murty: Thank you, boss. Let's see. And then there's a proposal I think from Chang, this is around improvisation. And so I think this one's also good. I don't think we have any actions on this one at this point Or I don't think changes on the committee. So, it + +Artur Troian: Yeah, blue fuse mean that proposal. And I think it's actually passed. + +Anil Murty: Great awesome. + +Anil Murty: Great perfectly fine and keep up with that. Sounds good. Our cash network Kepler, Mobile wallet. This would be nice. + +Anil Murty: Okay, so it looks like the action here is, we're just waiting on the team that we work with Wright Balls. + +Anil Murty: As far as Kepler Mobile wanted support goes. Okay. + +Boz Menzalji: Yeah, it's a Leap supports us Kepler doesn't for now no actionable items here. + +Anil Murty: And then actually haven't looked at these. Two we go. + +Anil Murty: Yeah, this could be a potential. Partnership on the AI front. Looks like we should have a conversation with these folks and see what this is about unless anybody else already has reviewed this and wants to share. + +Boz Menzalji: I'll reach out to wake up. + +Anil Murty: Thank you guys. And I'll review it as well. See if there's anything that would make sense for us to partner on A cash delegation dashboard. + +Anil Murty: Is this something that we already have? kind of, + +Anil Murty: Within Thomas? Or something doesn't look like it but it's wondering. + +Anil Murty: so, it was + +Boz Menzalji: Is the person who made the proposal here to ask a little bit more deeper questions on this. + +00:45:00 + +Boz Menzalji: I guess I'll respond and Learn a little bit more. + +Anil Murty: Thank you. + +Anil Murty: I think that's all the new discussion threads out here. Anybody want to talk about anything else, anybody on the call that wants to discuss their previous proposal. It looks like Max is saying they considered making a delegation dashboard at some point, but decided it wasn't worth the time kind of makes sense. I think. Yeah exactly. And as Huey said mint scan cows a lot of that. So that's kind of my reaction as well. + +Anil Murty: Cool, awesome. There's nothing else. Agreed. I think Greg also posted about that means canto, really cool. + +Anil Murty: I think that's pretty much any other open questions or things that people want to Reply to the community as a whole. + +Greg Osuri: No, I think it's awesome. So next time we'll be meeting. It'll be post Main Netflix. super exciting. It is our biggest update ever. + +Greg Osuri: I am without a doubt. Most important update. This is the kind of update that says, Akash apart. This is the first time you're actually diverging from what a cloud looks like, right from a decentralized version of Akash AWS to a version that AWS cannot offer so GPUs especially 100s is something that exclusive to Akash, So that's very, very critical. So I'm super excited for the update and hopefully the next updates are going to be in this arena of a different. Super cloud. That's not Amazon. It's Akash, right? It's up identity. That's being minted here. + +Anil Murty: 100% here, could degree more and I also think, overall, everybody's done, an amazing job of creating this ramp leading up to mainnet where we demonstrated GPU capabilities. First, deploying existing a models we showed taking in brand new am or CJ from the community was able to deploy sdxl. Literally a few hours after it was released yesterday. That was amazing. And then if we can next figure out getting the latest and greatest GPUs and then getting a foundational model trained after the meeting upgrade, that will be perfect ramp and launching GPUs. super awesome. Thanks everyone for all the amazing world. And with that, I'll stop sharing and hand things back to Tyler. + +Tyler Wright: Excellent, thank is there anything else that this is part where we talked about? Again, Any open discussion items? I know that they have mentioned gave people opportunities. Anything else that hasn't been covered than anyone wants to cover all. We have a couple of minutes left. + +Tyler Wright: again the notes will be available soon as well as a recording and the transcript as Anil and others have said, Thank you all to those that participated in GPU the test nets those that are helping with reviewing submissions and everybody that's building and supporting the cost community. Must appreciate all of you all, please feel free to anything and discord, go ahead, Andrew. I know you had a question. + +Andrew Gnatyuk: Yeah, just one question from the recent news about the passage six millions funded by their cashnet. I just want to ask if this was secured by the cashnet or specifically by Overclock Labs was the secured by + +Greg Osuri: It's good questions. Overclock Labs. Not a car. I don't know why they used a cash branding. I didn't. Really have a review opportunity for them before they announced this. but, across the protocol, it doesn't fund any other protocols or, That doesn't go through governance, right? So cash, only funds. Or the Treasury on your funds, The public goods and It's overclocked investment but I suppose they just use a cash branding. But if you actually go through the announcement. when I have a court right. Clearly says, I'm the CEO for our creators of cash, but that's a mistake on their part. + +00:50:00 + +Andrew Gnatyuk: Yep, got it. Thanks + +Greg Osuri: I just popular. So people are going to use Akash in ways that you don't approve. + +Andrew Gnatyuk: Yeah, so yeah, I know this is the public college recorded, but if there is maybe some Intel on the, passage will have some public talking sale. And I don't know but if do they have the airdrop or We'll have to, get more Akt for but some air drop for the talking holders. I don't really. + +Greg Osuri: It has plans. I don't know anything about it. I mean it says something about it again. the funding also really was people like personal funding me and Chang and Adam and whatnot. We funded this and Overclock only ever a little bit. not really much to add more transparency to and this deal was done here it was not in our latest, right? So it's like one of those things. + +Andrew Gnatyuk: I got it. + +Greg Osuri: The project is incredible. They've done quite a lot of work, I don't know if you saw the video. So there are also going to be a big user for a cash GPU. So it's gonna be a good sort of bilateral sort of just but they did say they committed dropping some talking to a cash holders. We don't know how much and + +Andrew Gnatyuk: Okay, yeah, thank you. + +Greg Osuri: Nice. + +Tyler Wright: Andrew, for that question. Thank you Greg for clarifying. Is there anything else with the couple minutes we have left? + +Tyler Wright: much, appreciate those, that will be listening or watching later. I hope everyone has a great rest of the day again. Sig. Economics will be happening next week as well as sick chain, six support and many others. And again, as a previously mentioned, I believe there's a working group session around Zealy which is previously through three happening tomorrow, Friday. And then look out for a working group around Docs 2.0, happening Monday and Adam Wesley will have an announcement because we're working group session for the HACKATHON before next week. But much appreciate Everyone have a great rest of the day. Thank you. All + +Anil Murty: Thanks everyone. + +Tyler Wright: right. + +Meeting ended after 00:55:48 👋 +