Skip to content

Latest commit

 

History

History
453 lines (266 loc) · 61.5 KB

008-2023-08-30.md

File metadata and controls

453 lines (266 loc) · 61.5 KB

Akash Network - Steering Committee - Meeting #8

Agenda

  • Review / discuss / address any community feedback, grievances, or conflicts.
  • Review projects list, prioritize / add / remove if necessary.
  • A look at open discussions in Github Discussions.
  • Open up to any other questions from the community.

Meeting Details

Committee

  • Adam Bozanich, Overclock Labs
  • Anil Murty, Overclock Labs
  • Artur Troian, Overclock Labs
  • Greg Osuri, Overclock Labs
  • Scott Caruthers, Overclock Labs

Participants

  • Andrew Gnatyuk
  • Andrew Mello
  • Anil Murty
  • Artur Troian
  • Boz Menzalji
  • Cheng Wang
  • Christian Cora
  • Denis Lelic
  • Emmanuel Arthur
  • Greg Osuri
  • Jack Williams
  • Jason Larson
  • Jeff Perry
  • Jigar Patel
  • Joao Luna
  • John DiBernardi
  • Julius Haus
  • karim guerdane
  • Logan Cerkovnik
  • Mark Ntim
  • Mark Sierra
  • Max
  • Maxime Cyr
  • Quentin Janninelle
  • read.ai meeting notes
  • Robert Del Rey
  • Rodrigo Rochin
  • Ryan Moorman
  • Scott Carruthers
  • Scott Hewitson
  • SGC | DCnorse
  • Stan
  • Tyler Wright
  • Zach Horn
  • Zeke E

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 were no grievances in the community at this point.
  • Anil talked about the project list.
  • Cheng talked about the utility of AKT. AKT is always going to be front and center.
  • Mainnet 6 tomorrow. Please look out for announcements.

Projects:

  • Mainnet 6 is coming tomorrow. GPU will be coming to Mainnet tomorrow.
  • Multi-currency support is also apart of the upgrade.
  • Econonics 2.0 is moving forward, because take rates are also being apart of the mainnet upgrade.
  • Anil mentioned that the Core team is trying to bootstrap demand. There may be specific groups that need to be spun up depending on audiences and need.
  • Anil also mentioned that the Core team is working on supporting datacenters as they bring on supply.
  • Anil talked about documentation for developers, as well as docs 2.0 that Luna and his team have been working on.
  • Anil talked about the on chain analytics. The indexer developed by Cloudmos team will used by the community.
  • Max mentioned that he and his team are working hard to get ready for the Mainnet 6 upgrade.
  • Cloudmos team has been working on the provider attributes. This was talked about during sig-clients and sig-providers meetings this month.
  • AuthZ is being worked on by the Cloudmos team, as well. There is n UI with AuthZ right now. Max mentioned that the changes for AuthZ are mostly done. They are waiting for updates from AkashJS before implementing.
  • Andrew asked "do we have aprox. number of A100 and H100 we are going to have on launch?". Anil answered that there will be a limited amount of A100s and h100s on the network to bring on demand. When there is more demand, there is the possibility to bring on more chips. Even datacenters have lead times on high performance GPU chips.
  • Greg reiterated that the approach is for supply is going to be more focused. The strategy is to build the incentive models AFTER the mainnet 6 launch.
  • The ramp up of high performance GPU supply will ramp up over time.
  • Andrew asked if the specific datacenters will be named? Greg mentioned that there will be events and a content push around bringing specific GPU
  • Andrew Mello mentioned that he has had DMCA complaints. He has been working on content moderation. Andrew Mello added terms of use to his providers on mainnet.
  • There is a provider chapperone in github for monitoring.
  • Greg mentioned that there needs to be a formal tracking. Anil mentioned that there has been a working group for content moderation since end of 2022. Andrew Mello attended many of those working group sessions.
  • Jigar and the Praetor team have started working on moderation API. The number 1 priority after the Mainnet 6 upgrade.
  • Anil talked about client libraries. Anil created a table that outlines the difference and similarities between Cloudmos and Console clients.
  • Max C has been helping Joe in making changes to AkashJS for both Cloudmos and Console.
  • Go Client library is highest priority for client libraries. Luna and Artur will talk about this after the Mainnet 6 upgrade.

Github Discussions:

  • The group discussed https://github.com/orgs/akash-network/discussions/247. This proposal is well written and make sense. Anil wants to drive demand on the network. This proposal looks to drive demand on the network.

    • Julius asked "Is this snap proposal connected with the announcement of Leap Wallet to also work on mm snap for cosmos chains or is this from a different team?" John mentioned that the products are different. John wants to give ethereum users a seamless way to use Cosmos.
    • John mentioned that Akash Network is an example of something that is in Cosmos and not in Ethereum.
    • This proposal makes the Akash experience optimized for metamask users.
    • Greg mentioned that there are 20 million users on Metamask.
    • The biggest drop off during Akash Console is when users have to install the wallet. With metamask enabled, user experience will improve.
    • Metamask has swapping capabilities in wallet.
    • There will be a gas fee abstraction level. Users can pay gas in whatever token they want. A wallet is the biggest issue with Cosmos.

  • The group discussed https://github.com/orgs/akash-network/discussions/299. This proposal is well written,

    • Anil talked about driving demand as the highest priority.
    • Greg's thesis is that the website is something that is hard to quantify. The website is a key touchpoint, and the existing website is hard to maintain.
    • Tyler mentioned that Piyush and his team has come to working group meetings, participated on Discord, have been very responsive, and the team is very capable.

  • The group talked about https://github.com/orgs/akash-network/discussions/300.

    • Greg mentioned that some people believe that decentralized networks can NOT do training.
    • This project is set out to show that decentralized infrastructure can do training.
    • This will be a proof point. This will also establish Akash as a brand in the AI space. This model will be published on Hugging Face.
    • Logan, who is on the call, proposed this to Anil during the GPU AI Incentived Testnet.
    • Logan is part of Thumper.ai and they want to collaborate with Akash to build a model that can be used by a wide range of folks.

  • https://github.com/orgs/akash-network/discussions/298 will talk about this next steering committee meeting.

Open Discussion:

  • Cheng and a member of the community were talking about the utility of the AKT token in chat.
  • Andrew Mello mentioned that demand can grow by focusing on those that care about cost savings.

Action Items:

  • John and his team to move forward with Metamask proposal.
  • Akash community to prioritize content moderation after the Mainnet Upgrade.
  • Andrew Mello will make a PR for his provider chaperone work.

Transcript

This editable transcript was computer generated and might contain errors. People can also change the text after it was created.

Tyler Wright: Alright, can everyone hear me? All Welcome everybody to the steering com Monthly meeting number eight. During the steering committee meetings. We go over the projects list as well as github discussions, have open discussions and do what general recap of all things that are happening in the various special interest groups and working groups. Usually Anil walks us through the agenda and I think he will continue to do. So today I know that Neil has been feeling very sick again just generally and there's been a lot of work that's been happening across the community with the core team as we prep for the main net, six upgrade that's happening tomorrow. I'm going to throw.

Tyler Wright: Countdown for the block height in the chat for everybody. But again, there's documentation available as well as a number of announcements that have been shared and discord and other channels for folks, if they have to date on all things related to the meeting, that's the upgrade. It was also a six support meeting that just ended where again Some other folks from the core team. Talk about the main net upgrade from other issues that will be high priority after the maintenance except grade has completed. But again, if you want to track any of those other conversations in August 6 are working groups. Feel free to look at the Readme files. Attached to the respective figure working group, you can see notes transcript. And find recordings on some of those meetings. Without further ado, I'll hand it over to a Neil The Denis kicked off with the typical agenda for the steering committee. I'll take notes in the background.

Anil Murty: Thanks Everyone, as Tyler said I'm pretty sick so I am sorry about if my voice isn't coming super clear and if it's not clear at some point, just let me know. We can probably have Tyler do this instead of me But yeah, that's a good and share my screen. And we can just walk through the agenda, like we usually do.

Anil Murty: All right, So Once again, thanks for joining us and I know we have a few new people on the call today. So for people that are joining us for the first time, just a quick recap or quick overview of what we do on these calls, the steering committee call is typically a once a month, call, where we go over the general sort of updates that are happening across the board. I cross all the things in working groups at a high level and then we also discuss any sort of issues or conflicts or things that need to be addressed from a community perspective. we review the list of projects that we have and if needed be prioritize, remove or add anything on that list,

Anil Murty: We then go ahead and review some of the new discussions that have been posted to the Github discussions board. Since the last time we met here and if possible we review some of the comments there discuss and potentially make decisions about whether we want to be supporting some of those things or need more information or outright doesn't make sense. And then at least we open it up to any questions from anybody in the audience here. Before we get going on these items, I just want to check and see if anybody has any top of the mind things that they want to bring up before we get into.

Anil Murty: Cool, any sort of community feedback from the way things have been running in the CIGS in the work groups? Any sort of conflicts that we need to talk about that. Anybody can think of

Anil Murty: I'll take that as a no, not that. We have had any in the past. I'm going to move on to the next bullet, but if any point at any point, somebody has any things they want to bring up. We'll have an opportunity to do that at the end of the call as well. So with that, we'll move on to looking at the list of projects. So here is the page that has the list of projects that are in progress right now. So top of the mind obviously is GPU support so unless you've been living under a rocky, probably know that we have a mean it upgrade schedule for tomorrow. That's going to Bring along with it, a whole bunch of things not just GPU support, but the big one among it is going to be obviously support for cheap, use on So that's It's marked as Beta but soon it's going to be complete. Hopefully, the made it upgrade should go. tomorrow. So that's the first time here.

Anil Murty: The second one is table payments and settlement. So really this item honestly is multi-currency support which is also included in this upgrade so that code changes for that are already being tested. And they're going to be part of this book as well. And then the Economics 2.0 proposal kind of fits in that same bucket. Which it's the one that introduces the idea of take rates in terms of being able to get a certain percentage of the fees that are paid or the network and put them into a community pool. So all three of these things are going to be part of that mean a great tomorrow. We saw questions on any of that from the community or does anybody else on the call? Want to add any more context and anything that's going on in that.

00:05:00

Anil Murty: Cool, Awesome. The only other thing I want to mention sort of on that topic is one of the things that we've been actively doing, which doesn't actually have a specific project associated with it, but We have been actively, who generate demand from a tenant perspective. So, time talking to a whole bunch of AI startups, as well, as independent developers building machine learning and AI algorithms and seeing, how they can utilize Akash in regardless of whether they're read three developers or read through developers, trying to get them to use Akash, as we get to mean it, and we'll see if there is specific, work groups that need to get spun out of any other effort to build specific features for these types of users. But for now, it's, just in action in a activity that happens sort of the background.

Anil Murty: The other side of that equation, there's been a very concerted effort on the supply side of this, to make sure that we have enough GPUs of different types available on Mainnet when we launch. So, one of the activities that has been going on driven by the OVERCLOCK team, but with a lot of support from the community, the plateau team and others. Is sort of doing two things. One is making sure that if not all of the providers that GPU clusters on the testnet on the inside. End up setting up a cluster on the main it as well. So, I know Tyler, you've been involved a lot with that effort and at this point he have, I think most if not all of the testnet providers signed up to support and spin up providers on the minute, right? Tyler

Tyler Wright: That is correct. Yes.

Anil Murty: Yeah, so, we're looking really good from that perspective to have, a decent number of GPU providers on Mainnet, At the launch. and in addition to that, we have communications in conversations going on with three or four new data center operators, Who we are directly partnering with to bring the most high-end GPU models including the H100s, the a180 gigabyte and a 100 40 gigabyte chipsets onto the network. So I think it's coming through together really well in terms of launch plan and then there's a few things that I'll talk about when we get into the discussions area of this meeting as well, that will add to the launch portion of it. Moving on down through the list and please interrupt me at any point if there's other things to add in the things that we're talking about,

Anil Murty: Certification program, Even planning and Akash Hackathon. These are primarily things that are being driven by I think, Adam Bosni still out at Burning This week, I don't know Tyler or anybody else on the call, if you're able to give an update on any of these items.

Cheng Wang: I can give a little bit of update here. So as far as a certification program that is temporarily on pause. For the moment as we continue to, focus, our resources and attention elsewhere, within the community particular about endurance planning. And, of course, the HACKATHON I think, which is well underway as far as signups go.

Anil Murty: And I know at least a couple of folks hit me up and we're meeting with one of the teams that is working on some kind of an ML deployment tool and actually this team was on the last call in the last time. I think talking about it as well. This is the Agora Labstein. So we're meeting with them next week to see kind of an early version of Fort where they're going with their HACKATHON idea. So, thanks Appreciate that Dockside Migration. I believe this is at the stage where it's getting ready to get pushed to, some sort of publishing Location. Right title.

00:10:00

Tyler Wright: Yes, there was a safe documentation meeting last week where Luna and Andre worked on next steps for the handoff. So that again, docs 2.0 can start to be reviewed by folks in the inside of vanguards community that will start to happen after the main end upgrade. So, they'll be more announcements of policy and I know that there's some other folks that are interested in supporting documentation, some other items with the website. So they'll be further conversation around that after the main ended upgrade.

Anil Murty: very nice. Thanks to Luna 19 for all the work. They've done on this. Go.

Tyler Wright: Absolutely, I just want to make a quick call out about the events and hack upon as well. I know that the next events coming up are going to be Massari main net and then the car obviously Cosmo versus coming up shortly. There's some folks on this call, I believe that are going to be participating in those events along with members of the core team. So, again, look out for announcements and then

Anil Murty: Thanks ito, appreciate that additional context on the documentation item here. The other thing I think is worth mentioning is so this Migration is primarily something that being driven or this item specifically, something that was led initially driven by the Luna and Quasar team. And the intention here was to take the existing content that exists on the existing oxide reformatted in. A way that is more appealing to developers and improved looking feel of it as well and launch it on potentially a new platform.

Anil Murty: Outside of that, we also have an effort and this is, I think at this point primarily driven by the Overclock core team, Scott characters as well as Arthur on our side and this effort is basically to focus primarily on development contribution documentation. So one of the challenges we have had is you've had a lot of success with people contributing on the client side, but not to the same level. When it comes to the core products, meaning on the go side, go go lang, develop inside. so, one of the things that we're going to look at postmen and upgrade is,

Anil Murty: Bringing in really good documentation for people to be able to get started on contributing within the go code basis. we still have to figure out where the documentation exactly is going to live. It might be a combination of the Github. Readme's MD files there as well as the dockside. So something to think about

Anil Murty: The next one, this is on chain and lyrics. I think we've talked about this at length in the past in the last steering commit Correct me from wrong, Tyler. But I think that we communicate the decision that now, we're gonna basically adopt a Cloud Moses indexer as a default for our analytics needs given that cloudmosis now merged with Overclock Labs and Max in team, both actually have been working on a whole set of things that they need to do post-mated upgrade to make sure any of the changes that have happened. In terms of the manifests in APIs, that have changed as a result of the main upgrade. Those are taking care of in not just the deployment tool but also the index. So thank you so much team for staying on top of that as well. So hopefully,

Anil Murty: But there might be a slight hiccup post-mated upgrade for a few days with the indexed data, but that should be resolved in a few days. I don't know if there's anything else you would like to add Max or able from the club hosting

Max: Hey guys, not much. I think you've said it We're working very hard to be up to date with the mainnet tomorrow and have as a low downtime as possible.

Anil Murty: Thank you so much, guys. All right, moving Provide attributes.

Anil Murty: This one we talked about this. I think during the provider I would say to provide a SIG. I don't think we talked about it in the last steering committee, but The Cloud most team has done work on this side as well. So, one of the things they did pretty much, the same week as when the MERGIO is announced is they put out an update to Most Deploy tool where if you go in with your provider wallet, you can now edit attributes in cloudburst, apply and the intention with this again, is not to become a provider management tool like Predator, that's a completely different use case. The primary objectives here were to be able to test what sort of keys values work for provide attributes, and offer providers, and easy way to be able to change them within the UI. So that's implemented in again, thanks so much for most team doing that.

00:15:00

Anil Murty: The next one here is auto spend management. So there is a spec for this, if you haven't looked at it, it's linked right here. But essentially, this is the spec for being able to delegate funds to somebody else to use for deployments. So the ultimate use case for this is Like you have with traditional clouds when you go sign up at GCP or AWS or one of these services, typically if you're a brand new user, you get, $100 or $300 or something to use on cloud credits, what we ultimately want to get to is a user experience similar to that. Right now, it's sort of would say halfway there where

Anil Murty: You can delegate funds using cloud most, and you can use funds delegated to your wallet and you deploy, there's a few challenges with it. One is for the person, that is delegating the funds. So for example if it's the finance team on the overclock side or marketing team, running on campaign or something.

Anil Murty: There is limited or no visibility into, who the funds have been delegated to from the UI. There's no management of all the wallets that are there as well as information about how those funds are being utilized, how much we've just been utilized, who has actually utilized the funds potentially to deploy stuff, right? And then, the other challenge with authorized Spend has been that. In order for you to delegate Akt, to someone, they need to have a little amount of educati already in their wallet in order to receive the agari because I got to pay for gas queens. So solving that problem and Max see on the cloud, most team has been tackling both of these issues as documented in the stack and I believe either has already an update out there or is going to have an It pretty soon. I don't know if Maxis or the call and can providers any more information.

Anil Murty: Yeah, maybe not. But yeah, I think we're pretty close to having a push to cloud most that'll support more features. Go and Max.

Anil Murty: Yeah, so this is super exciting and I think Cheng and the marketing team will really appreciate this when it's out there because it's been a pain point. in the past when we have done, for example, the Carver campaign managing the wallets in, doing it all had to be done for the command line and there's no way to visualize it. And then of course, removing that fiction of having to have people have searched for Vacaty, already would go a long way in driving it option as well. So thank you for doing that. And since you mentioned akashias in that context, I guess I don't have a specific item here. I guess we can talk about client library, hold up for that. So that's authorized spend any questions so far or any of the topics and I'm going pretty fast through the list.

Anil Murty: It's policy of the second seat, is any questions?

Tyler Wright: Andrew from the chat had a question when you were talking about supply of GPUs and some of those data center relationships, who talking about ask a question, approximate number of a 100s and 100s going to have online.

Anil Murty: Yeah, great question. So it's a bit of a chicken and egg problem as you can imagine. on the one hand, we would like to have Infinite amount of supply, but in the second hand, we don't want to commit to a whole lot of, funds for that as well. So we're taking sort of cautious approach where we're committing to a small amount initially to start with and we're also and, maybe Cheng can talk about this a little bit as well in the context of marketing. It also going to be working on a demand in page for, attracting developers. And so, what we're going to do is we're gonna have a limited amount of a 100% h100. It might be, I would say somewhere around the

00:20:00

Anil Murty: yeah, in the tens of order of tens, not hundreds and that will be enough for us to be able to seed any demand that comes through. And then as we see the demand coming through, we'll be able to go and get more chips. It's from these relationships that we have. The important thing that we're doing is, we're forming relationships, with multiple data center, companies that have These chipsets available. And another portion of this is and it's both a good thing and a bad thing. Many of these data center companies themselves.

Anil Murty: Have a little bit of a lead time in terms of securing these tips and setting up the clusters with them. So for example if you want a 8 or 16 cluster 100s of h100 today and you go tell one of these data center companies I want an EDX cluster. They're probably tell you it's going to take you a month and a half before they can have that for you. So in a way it's a bad thing but it's also a good thing for us because it means that we can sort of look at what the demand is going to be once we launch and then go ahead and request him but yeah hopefully that answers your question Andrew there's any other follow-ups.

Greg Osuri: To add more light to Anil. Just said the

Greg Osuri: the approach we're taking is more cautious and more measured in Cyril going guns out here we have all year 100s in a pipeline. We have thousands of 100, right? Instead of, putting all those supply on Akash and Not Catch up to the supply and essentially having utilization. Remember, these chips are extremely high demand, right? So

Greg Osuri: Instead of having utilization, we are ramping up the supply based on utilization, So that's not going to prevent anybody from listing on akash, but considering there are no Online centers in the sense Hey you only get paid If there is a demand for your application, there is no on change centers yet. And that was on purpose, We decided to work on implementation of Incentives. Post GPA launch and not pre GPA Lodge because we don't want to go the same way or have the helium or file coin, right? These are failed and send a models. so, ramping up the supply in proportion to the demand. An acceptable level of utilization is the strategy we're going.

Greg Osuri: and so what that means is, yes, it's going to be like me tense, maybe hundreds in the first quarter till you go you Q4 and it could be some in the mid hundreds in Q1, And thousands in YouTube. This is going to be a very good ramp up. Although we have enormous. Qualified demand in the sense being in testnet, people were able to experience the product. People said People the way the product functions, but there's no way to guarantee unless they use a man, There's no way to guarantee that. Hey, this user is going to come use the product with our remain at launch. So it's a very delicate balance and balance need to be maintained with precision in order to be successful. So answer your question. Yeah, it's gonna be like 10.

Greg Osuri: May not be hundreds. At least. maybe hundreds my end of the year or it is X quarter, but it's going to be a slow ramp up based on Demand.

Anil Murty: Cool, thank you for the additional context. Greg it's really helpful. and by the way, yeah.

Greg Osuri: Will be shared a public. I mean Akash everything in the broadcast is public and…

Anil Murty: Yeah.

Greg Osuri: you can go to Cloud Moss and see how many GPUs there are. I don't think It's going to be any. Hidden aspects of Akash ever. This meeting as we, speak is recorded and published on RV. The notes of which will be transcribed and easily Googleable. And summaries of what I just said will be also available github if you want to go back and So every aspect of a car is public including the staff, including what we just talked about,

Anil Murty: Hundred percent. And just to throw an addition piece of context Andrew is, I don't know if you follow the announcements from the public cloud providers, but I believe even GCP. Google Cloud doesn't have h100 this point, and I don't think they've announced a specific date when they will. So and…

00:25:00

Greg Osuri: Nobody has h100.

Anil Murty: then, so just having, even a few would go a long way in terms of being able to, Position as well. And I see Logan, nodding his head.

Anil Murty: The next one here is content moderate. sorry I think Andrew you had a hand up right? Go ahead.

Andrew Gnatyuk: Sorry, about my last question. So what I meant is Will this be for example or right on Twitter for example, we are securing dance or hundreds more or where h100 GPUs from that data center. the community know where are they coming for? not just like they just releasing on the marketplace but that hey guys tomorrow, we're going to have more like h100 GPUs. If you want it, it's gonna be from Slovenia or Germany Data Center, Yeah, yeah, something like that. Like that info.

Greg Osuri: Yeah, most likely will be talking about a providers like we normally do. We probably will, do spaces and do events to introduce the providers to the community, So are we going to announce? I mean, depending on depending on the provider and depending on The context in size. Yeah, it's probably going to be some Definitely content. Push out there. yeah.

Andrew Gnatyuk: Yeah, got it. Thanks.

Anil Murty: Yeah, that's a really good question,…

Greg Osuri: And as you can,…

Anil Murty: sorry. But

Greg Osuri: I'm pretty active on my twitter, a little too active. Then I should be. So,

Greg Osuri: So, you can expect and expect some announcements.

Anil Murty: And Andrew that's a really good question. One of the things that I and bars and a few of us are really interested. In is actually forming closer relationships with a lot of these GPU data center companies, and probing for whether there is a potential for us being, some sort of a distribution channel for them, And if there's sort of incentives that they can offer us for doing that. And so that's something that we're pursuing actively with multiple providers and if something like that would work out then it'll probably Naturally lend itself to us talking about these providers publicly and potentially putting them on a website and stuff like that as well.

Anil Murty: If that makes sense.

Anil Murty: Also. all right, if there's no other questions on all of that. One more hand. Andrew Mello.

Andrew Mello: Hey I see we were on content moderation there. So I did just have a notes that I wanted to mention with the group here since we have everyone on the call. I, have been running some of the largest providers on the network now for some of the longest time and I am. Now starting to come into issues with DMCA complaints and other. Basically, mischievous usage of my providers. That, again, this now has become an actual issue for me because I have six Dmca complaints against one ISP in particular. So this is caused me to now, escalate this need to do content moderation and So what that means in totality is first of all, defining, what we're moderating, and as most providers, and

Andrew Mello: And companies do that's defined through a terms of service and an acceptable use policy. So to get that going, as I said, I am in urgent response mode. not waiting around for it to happen, but it has happened. I need to do something about it mode. I've gone ahead and created a terms of service and also acceptable use policy. I put that on my status pages of my providers so that they're publicly available for anyone that was to deploy on my providers. And then I additionally posted those

Andrew Mello: Into the Caution Ciders channel just a couple of days ago. Additionally we do have a github for this provider chaperone was the name of the GitHub that we're going to publish, some of this code under, but I am now ready to publish the code that I am using to actively basically enforce the acceptable use policy. So, again without enforcement without a tool running on the cluster to monitor for parents magnets ebooks, All the kind of things, that are gonna get you dmca complaints or worse, that's kind of now. Again like I said, I wanted to give everyone an update because it's been moved from kind of just sitting on the back burner to this problem needs to be salt. Now and as we continue to grow, this problem will besiege, all providers at some time or another. So I want to be very transparent publish as much as they can around this and I will go

00:30:00

Andrew Mello: Depth on our next content. Moderation working group, but just wanted to rate some attention to that this is now something that, needs to be solved for and I am looking to publish and help document along the way.

Greg Osuri: Yeah, thanks for that.

Anil Murty: Andrew.

Greg Osuri: I think it's important to create a project now for that and start tracking as a project, the progress on this effort, It's a good point that he brought up Andrew. It's been a top of my mind and a lot of the community members as So yeah, that's doing more formal tracking of this initiative.

Anil Murty: Hey, Andrew and Greg. So just to be clear Andrew and I don't know if you have joined any of these meetings and I don't if you're seeing the screen right now and we will actually just go to jump into content moderation but we have had a working group for content, moderation for a while. Now formed in December and we have had a few meetings on them. And this specs for specifically, what, we want to achieve with content moderation and to APIs defined as well. I believe it, one of these meetings, one of the action items, there was for Adam Bozinage to create a repository for you to check in the code that you had shown as one of these points.

Anil Murty: I think we may have lost track of that at some point and I don't know if that repository ever got created. I will go and check if that repository got created and if it does didn't I'll create it or have asked somebody to create it for you so you can check in your code. But yeah,…

Andrew Mello: Great. Thank you.

Anil Murty: yeah, thanks for bringing it up. But just to be clear to everyone, it's not like we have ignored content. Moderation. We have had a working group on. This is actually one of the first working groups. if not the first working group we found as soon as we decided to go to this community model, I mean You right here. December 22nd is when the first meeting happened and we have to find the specs and the goal here was, the Plato team was going to work on one portion of these APIs. Obviously, the Plato team has been extremely busy supporting all sorts of other things since then, but the intention is for them to switch back to working on which was it The moderation API I believe is what they were going to work on.

Anil Murty: As soon as the minute, I mean it is done. But yeah, point taken Andrew, we will continue to push steam on this and literally I was just telling Greg last week, I think that we're going to put gas on this, the movement we get ended up Go ahead.

Andrew Mello: Just my last point there is really that again, through this process, I learned what was missing and what we missed in our SIG. There was the terms of service acceptable, use policy and I think we kind of need to get to a point where we provide some kind of template boilerplate to providers again. that's kind of the discovery here. So again, we'll get into it on the next one, but just want to bring that up that I'm giving it special attention because it is now an actual problem. Yeah.

Anil Murty: It wasn't Andrew, I appreciate that integral. I just want to make one point before I hand it to So you mentioned that after you created this term of service you shared what you did on the insiders channel, Can I ask that when you have such things to share by all means share them and discord but also please create a pull request in this repository or, if it's for something else creative request so it actually gets a formal review and it's in there and it's not lost in the stream of data that flows through discord all the time. Yeah, so please create a pool request for just,…

Andrew Mello: Okay.

Anil Murty: create an MD file here if it's a code change somewhere then we can figure out which if we need to create new repository for it. But if it's a spec or it turns the service or something that you want to call out, please create a pull request for this and open it up. So, a few of us can review it community, can be aware of what's going on.

Andrew Mello: yep. I'll put the templates in there and, we can just use that to work from. As I said, though, it's gotten to the point where, there has to be something in writing and there has to be accountability for things because, an ISP Cutting me off for DMCA complaints is not an acceptable, use if we're gonna deal with it. Yeah.

Anil Murty: Nope. Not arguing with you on that at all. Thanks so much. Hey, go ahead, Jigar.

Jigar Patel: Yeah, just wanted to read it. We have started working on. It is just because we were very busy with the GPU or testnet, and ready for upgrades. And then after the press, obviously, we gonna go back and that's our number one priority.

00:35:00

Anil Murty: Awesome. Thank you, Jigar.

Anil Murty: I guess that color is Moderation. Any other questions or things? People want to bring up?

Anil Murty: Okay then moving on to the next one. Here is client libraries. Mainly, I want to talk about two things on this One is the last time we talked about Akash Gs and how we want to use Akashiases, the place where we document and keep basically the common set of

Anil Murty: Features and Functions Between Cloud Most and Console, Since then, I have gone in and created this table that calls out, specifically the areas of commonality between the two products regardless of whether they serve with three or web to users and where we can basically choose a specific implementation between the two products and abstracted out to akashias. So it's utilizable by both the clients without having to write the code or write custom implementations and it's also available to anybody else that wants to build similar clients. So a lot of the effort for this is actively going. I know, Max C has code changes right now in akashias. So if you actually go to Let's see. Cheers, you can probably see. There's been pull requests.

Anil Murty: From Joseph, as if, a few minutes ago, updating it to support the new beta, 3 API, and then Maxie may changes to fix issues in person storage as well as ipleases. just last week. So basically what we're doing here, as you can see is we are, abstracting out features that are common to both the clients into akashias as libraries that can be utilized by any other clients. And then cloud, Moses adopting, that common implementation if they have a custom implementation right now and making sure that none of this breaks any of the existing functionality, So thank you again Cloudmo's team and Joe for All the work on Any questions on akashias in what we're doing here on the client libraries front.

Anil Murty: Didn't take that as a no. The other thing on client libraries was, I was wanting to get an update on what is going on with the Java or go libraries. If anybody has any, I'd Luna in team or anybody else.

Anil Murty: I don't think I see Luna on the call.

Tyler Wright: Call. I know that at the last six clients meeting where I think Greg was on that, it was some discussion about the Go client library being the priority. I know that. Arthur. And Luna have previously been talking closely about Go client library and a solution. I think with the main debts pending that discussion, has been pushed aside until after Maynet. But I know that again go as high as priority for folks and then Java secondarily. So, I think there will be continued conversations after meeting up between Arthur Luna and other team members that want to talk about the ghostline library.

Anil Murty: Yeah, that makes sense. Thank you Tyler.

Anil Murty: And I think all of these other items, there's no active work going on right now to the best of my knowledge. So, I don't think we should spend the time going through each one of these. Are there things that are not on this list that you were expecting to talk about anybody from the call. Can you think of anything else that is being worked on within the six over groups? That is not documented in the table?

Anil Murty: I'll take that as a no and move on to the next topic on the agenda, which is reviewing all the github discussions that have come up since the last time we met as a group here. So just going down here and just looking at the dates when these were submitted June 20, Please start with these ones here. So the first one here is drinking Metamask to Akash via the Cosmo snap. I reviewed this one yesterday and I've been pretty late to getting to this one. all the questions that I had, I think were already asked in this. So the sounded really good to me and it's also pretty well written. So John, I think you're on the call. Thanks for doing such a good job of writing this up. and when I think about,

00:40:00

Anil Murty: Proposals that I feel like we should prioritize. I basically boil it down to do things in my head. If you look at where we are in terms of the network today, our primary objectives obviously should be to drive demand, and ideally drive demand from AI developers. Given how much attention that is receiving now and given that the AI workloads generally tend to have lower, five nine requirements compared to traditional Web services. So, driving demand for me, I'd just kind of the number one and then second and more broadly driving more demand from all sorts of users, right? And then the second is, of course, making sure there's enough supply and creating incentives for providers to have supply. So any proposals that I see that try to achieve one or both of those objectives automatically like our winners, in my mind in this definitely is hitting the first one. Pretty hard.

Anil Murty: So that's kind of my take on it. At least I just want to kind of see if anybody else has any thoughts or opinions on this that they want to share.

Anil Murty: Go to Andrew.

Andrew Mello: Yeah, in terms of, demand generation, I think that a particular branch that we have yet to explore is really our price advantage over traditional, cloud providers and, exploring what that looks like in terms of again, there's a certain subset of communities out there that are always focused on price and always looking for the cheapest compute the best price. And so I would propose that we also include continuing efforts to reach out to communities and ecosystems that are price sensitive and they're, quite a few of them and we bring a huge price advantage over a traditional container deployment or Kubernetes platform.

Anil Murty: Thanks I really appreciate that perspective, but just to be clear and I don't know if you're on your phone and not seeing the screen, but we are specifically talking about the proposal to bring a Metamask to Akash using the Cosmo snap. So I was just mentioning that this definitely helps us drive demand. So what I'm asking for, what we

Andrew Mello: Sorry to miss that. Yep, I did not see the screen there, sorry about that.

John DiBernardi: Yeah, I can see where Julius just said he said, that is the snap proposal connect with the announcement of lead wallet. No. So this is actually two different implementations. the one that we build at Mystic Labs is a much more holistic, it's meant to benefit the Cosmos ecosystem as a whole, sorry about that. I'm getting notifications on here. It's meant to benefit the cosmic ecosystem. It was built dynamic and in nature to really support every chain. And every coin, type with the sole purpose of funneling, Aetherium users into the ecosystem, leaps is more of. And I know I believe that just took something about the coin types, Matt thing that we built out which is cool, I'm very excited for them doing that as well So there's I believe is more so focused on improving the lead the wall experience. ours is giving a theory of users

John DiBernardi: Seamless method to access the cosmos and really it's highlights. And I do think that, Akash is something that has a unique. I was talking about this on a space the other day. I was like, what is really a unique driver for Cosmos? What is something that is unique, in this ecosystem, that is just not an ethereum or these other ecosystems. I think probably, one of the biggest examples of that and I think that, it should be right in their face. Hey, pop up a server right away. Using your mana mask wallet, no strings attached, one of the easiest things you can do. and I think the main basis of this

John DiBernardi: Also is not just because we can already support Akash right away, but it won't be, the full experience. What this proposal aims to do is really, add notification support and really make it as simple and easy to use a cash experience. So for example, your server is running low on balance. You want to be able to get that notification in while So you're okay. I can top it up right away. If that makes sense.

00:45:00

John DiBernardi: Hope I answered your questions though. Just randomly.

Greg Osuri: So, I think it's important to add about 20 million users on metamask. That will be open. And then after Akash, I think that alone is the incredible injection. As I don't know how our biggest drop off point when it comes to using a cash network. The console is the step where folks need to install the wallet. because typically takes, a good portion amount of time to install a wall and get some tokens. From, it cracking or something. It takes 10 to 15 minutes, right? but if you have Metamask enable any, metamask user can Actually use Akash without that steps about 20 million users.

Greg Osuri: And read a mask also, has swapping capability in wallet. Which is advantage To be able to get a three tokens or usdcs. So I think this a lot more benefits. To this proposal. Then, the game is incredibly high price. So, I mean, I think this, I completely supposed proposal and I think I would encourage community members to take a deeper looking into it and we ask yourself what? Values means to the ecosystem and it's obvious, I would highly encourage you to support as well.

John DiBernardi: Appreciate that very much. And I think there's actually another thing that's worth mentioning It was just announced yesterday about finding the future rpcs of this. That was so the Rpcs are now pretty much handled for the upcoming year but there was two aspects of that that I'd like to share that are very important. One of them is, a third of all IVC, swats revenues going to the ADAM Community pool. But I think the more important part right away is we're actually adding this gastly flexibility module, that we're going to build out. It's very similar to The Attraction module, that people implementing on chainsaw. But what essentially is it's gonna do is allow people to pay gas in whatever token they want.

John DiBernardi: And that is another barrier to entry that's being lowered. So now people are going to be able to pop up in a car server, and pay their gas and add them or whatever token they already have in their wallet. So now it's another piece that, It's less of a barrier for a very users because it is the bottom line, you can ask anybody that is doing research in between why a theory musics are not trying to cosmos not because the projects are great. We have some of the greatest texts and the greatest projects they just don't want to do wallet. And also getting the actual cosmos assets is a pain. Nobody wants to go to a separate website and bridge and All your different wallets. it's poor user experience. The less steps to reach the goal, the better. so, Just figured it was worth mentioning that.

Anil Murty: Thank you John for walking through that again. And so, I guess at this point, haven't heard anything negative about this, so I feel like it's pretty good to move to the next stage of this. So, If you need any help with shepherding, it next through on our side, myself bars, Tyler Cheng happy to help you out with that.

Anil Murty: in the interest of time, I'm going to move on to the next discussion thing here.

Anil Murty: Teach what we do. I'm gonna take this one next Three. So, this is a proposal by human HQ and I believe the person behind this is Piyush, who has been a very active contributor to the website recently and I believe also built a AI app for doing child TV, style stuff with Akash documentation. So amazing. Contributor, This proposal I think is super well Very detailed and has, pretty much all the things that you expect to see in a proposal that's written

Anil Murty: I reviewed it this morning and I put my comments in here. I don't know if other folks have had time to review it yet. I love this proposal overall. My only concern with this proposal is again, just goes back to sort of what I look at when I think about proposal submitted. The stage of where the network is. I think about, primarily what are the objective? We're trying to drive demand, or we're trying to drive supply or incentivize supply both ideally, and while In Theory of website, Does contributed driving demand. The thing I wonder is it's not like we don't have a website right now. The question is, Do we want to spend this much money, revamping the existing website?

00:50:00

Anil Murty: But if we think that revamping is going to significantly drive up demand for us, then it might make sense for suspended. And so, that's kind of what my comment here was. I'm curious if there's other people that have talked about this,

Greg Osuri: So website is one of those. Multi-faceted, Vectors that can be easily quantified in terms of either driver supply. It is a central stone for all aspects. Now, if you look at

Greg Osuri: I think what this proposal is trying to address is the challenges to maintain the website and to update the website has been great. because, the tech stack was very advanced or Not very common. And updating hasn't been the easiest, right? So think this proposal is going to do is it's going to make, Andrew capabilities to the website. They'll explore new areas as well as make it easy to maintain. Judging by the depth of the proposal and getting by pious and the teams prior work, pretty high confidence in the team that they can deliver the quality games work as well. So,

Greg Osuri: The question I was like we have a very thorough detailed proposal as to what they want to do. The benefits may not be quantifiably, measured at this stage, because websites are very hard to quantify with measure. but qualitative aspects, we have high conviction on in terms of maintainability as well as well as brand as well. A lot of aspects of Right. So there's clear qualitative advantages for this effort. so, even there doesn't fit in a single bucket, I think it addresses so many aspects of a caution network it's important to have you maintainable web. So that's my pieces and I think anyone has to build a distinct is very good. So how suppose proposal

Tyler Wright: I know we want to move on to a couple other proposals but I do want to second what Greg has said. I know this group has been very responsive and has been very much available over the last month has given a number of hours and his love, put their money where the mouth is and just delivered, a great deal of work.

Tyler Wright: Some of these pages are going to be brand new pages. I know that Greg did a great job of setting the tone with the website as it is today. But there are some pages that Greg didn't have the time to develop because he was working on a number of other things to the case of the business. So whether it be the ecosystem page, making improvements to token page, or even again, adding brand new pages that make it easier for folks to understand what Akash is. And then use Akash is all part of the efforts from this team and again, huge and it's also four other members that are gonna be focus on the website. He got a team of eight total that are going to that are very interested in a caution and work a number of things. But change to the website again It's peach posts for folks. So again I know want to move on. Anybody has any questions Feel free to get involved in the working group Akash website, discord or just respond to this.

Tyler Wright: Proposal and Github discussions and I know that Peters can answer and his team can answer any questions again, they follow their format of the open structure and been involved in meetings. And so whether it be the next working group of cosh website meeting, which is not going to be tomorrow for the following week, or again, in discord, folks can be transparent and talk to them because again, I think they're really quality team that we should embrace in the community.

Anil Murty: A good thanks Greg for that additional context.

Tyler Wright: People.

Anil Murty: Yeah, and again from my perspective, I just was asking this purely from use of funds perspective but no disagreement on the challenges we have on the current website. I've been personally involved into update it a few times.

00:55:00

Anil Murty: Any other parts of questions from anyone? On this proposal. Before we move to the next one.

Anil Murty: Let's see what else it got. This one and then I'll do the last one really quick. So this is a proposal I put out last yesterday, evening late evening so I don't expect everyone to have reviewed it already but I did get one comment at least from Derek. I don't know that it's on the call. I'm gonna check really quickly. I don't see him on the call but he did ask some really good questions. I responded to them, really quickly right before this call in the last few minutes, Before I was getting ready to get on the call, I haven't answered it fully but I wanted to see if there's any other questions before we dig into these ones. Since Derek is not on the call.

Greg Osuri: But it's what I did a little more context was proposal. I think this proposal To is very capable of creating a foundation model the reason for that is You see quite a lot of Conversations and Twitter and all over about the fact that decentralized networks cannot do training. And are only mental inference and client tuning. which is not rooted in reason, but I think it's Simplifying. The problem quite a lot. The effectiveness of training is proportional to the density of incident quality of compute.

Greg Osuri: And the latency of the computer, A car is a general purpose, computer network and a key feature of which is hyper granulars attribute selection, right? So, By demonstrating this granularity in filtering for the right type of resource. And given the general purpose capability, you can actually do a foundation model from scratch. So overall, I think it's a great proposal. give us that proof point that we can always reference for new users that are want to do training our cash second.

Greg Osuri: Establishes Akash as a cautious brand in the AI community, which we don't have much as of Now, we are known in Crypto community but from AI's standpoint, we're sort of unknown to the most part, right? So with this model being published a hugging phase, there is a REFEREN proven aspect of a cosh network can be used for training. So there is enormous advantage of this model and funding this model training our gosh. But yes let's get into the questions.

Anil Murty: Thanks And I totally forgot. I should have said before we even get in the questions Logan since you're also on the call thanks for joining. Logan actually was the person that proposed this idea to me several weeks ago as we were thinking or testnet. And then I was like whatever we first work on something that we can show in the testnet and so we together, worked on the benchmarking application that was used to benchmark all of our GPUs using Archbench and Jupiter notebook. So Logan, thanks so much for doing that work as well. I just wanted to give you a few minutes at least to talk about, anything that you want to talk about with this proposal.

Logan Cerkovnik: So thanks Greg in Anil, for this proposal, it's a collaboration between a thumper AI my startup in a caution what we're trying to do. is build a Creative, Commons Zero, License Foundation model, that's compatible with Diffusion one five open source software and so we think it's a great opportunity for us both the collaborate and create a unique model that's usable by a large range of open source, software tools out there. We're really excited about this opportunity and we've gotten some feedback already. We'd love to get some more feedback on that for quite different questions that people have. and I just say what we're really excited to let me know if anybody has any specific questions, or comments, or thoughts on this and we'll go from there.

Anil Murty: Thanks Logan. Yeah, so I was going to talk a little bit about this question from Derek, but since he's not on the call, I don't know if I want to hold everybody up doing Any other questions from anyone on this proposal? Before I hit the very last one on the list?

01:00:00

Anil Murty: I think that as a no and, Logan and I will follow up with Derek in the comments on this. So if you want to take a look at those after we post them, take those of us. And let's see what else is left. it discord documentation part. I've actually not read this one. I don't know if anybody else has and can talk about this if the person that grow it is on the call.

Anil Murty: yeah, so as I'm looking at this it seemed like this is somewhat in line with somebody efforts that have been going on both from, Posh as well as from Scott on our team to build essentially what we're calling Akash GPT or Cosmos DPD type of thing scenes. It's somewhat similar to that, but I'm not entirely. Sure but I guess if there's nobody, they can talk about this. We can contact to the next Student Committee meeting. All right. Let's do that. Any others that anybody wants to talk about, who's on the call. And has written a proposal that we haven't touched.

Anil Murty: All right, take that as a no. And then the very last thing we do is we just open it up again for any general questions from anyone, on the call about this meeting, any of the CIGS were groups, anything at all.

Anil Murty: Coaching.

Cheng Wang: Yeah, I think it's as worth mentioning kariman and I were just going back and forth a little bit on the utility of akit. It's a question. I definitely pops up with a pretty fair degree of consistency. So just want to take every opportunity, I can't emphasize that the utility of Akt will always be fun center, as far as any decision made in terms of multi-currency support, and as a vision Akt 2.0. And of course, in all of our economics design having, make it cheap fun centers, always going to be one of the key pillars here and so the take rate is one of probably the primary component for ensuring that all usage on the network. Regardless of currency, it creates back to create value for the caution at work. Not just the community, but also ultimately tell people there's as well.

Anil Murty: John, thank you Jane and I hope that answers your question, yeah, I think that wraps it up for us. I know it a little bit over time but honestly this is my favorite call of the month. And it's always nice to review everything that's going on as well as a whole bunch of people. So I really appreciate everybody taking time out of the day to come to these calls and with that, I'll stop presenting and things back to Tyler.

Tyler Wright: Excellent. Thank you. Anil.

Tyler Wright: again I appreciate everybody. Staying a little bit extra. Again, please. Look out for announcements over the next 24 hours at the pertains to the main. That's six upgrade. There is going to be documentation, that will be distributed for folks on the ballader side. Again, we're looking around one 12, I believe Eastern time tomorrow. For that Mainnet, Six upgrade. So, folks, again around maybe two hours ago, will hopefully be online on the ballot, either side to upgrade their infrastructure. And then again, there will be an announcement for folks that are providers when it is time to upgrade providers. Please don't upgrade providers before. That announcement is made available Be discord. It will be a number of places with documentation and instructions on how to upgrade. And then again, if you're going through the pray to our path, I know the page or path has been a play to our team has been updated on

Tyler Wright: Upgrade path and they're working on the upgrade path right now, for those that are using krator. But again, Please do not upgrade infrastructure. Until you're told by members of the Overclock Core team Again, much appreciate everybody's effort and participation at GPU Test and participation in very support channels As we get ready for this meeting, That's six upgrade. It's been a true team effort to get us here through community effort. So again, please everybody, mark your calendar sets. Arms for tomorrow midday so that we can reach consensus as quickly as possible in the validator side and then start to bring GPUs onto the mainnet.

Tyler Wright: But again, there will be a number of announcements coming soon. So if you have any questions in the meantime, please reach out on discord, to myself, Scott Carruthers, and some other folks in the community. out. And there will be a live stream tomorrow where we will open it up to the community, to watch the process of us, going to the main upgrade, so you'll see members of our core team potentially cursing or, just smiling and holding on while we reach consensus. But again, there will be some announcements on social media with a link to that YouTube live stream tomorrow. So you can kind of watch an experience upgrade alongside us.

01:05:00

Tyler Wright: the notes and the recording will be available soon. Much. Appreciate everybody's participation today again. Please feel free to get involved. In discussions on GitHub, join a SIG or working group. Just continue to get involved and continue to push, a cosh network forward. I hope everyone has a great day and we'll talk online. I'm sure. And at 6:00 tomorrow.

Anil Murty: Who thanks?

Tyler Wright: Thank you. Earth.

Denis Lelic: About.

Cheng Wang: Thank you, everyone.

Meeting ended after 01:34:28 👋