mmq

We have built similar abstractions directly on top of Kubernetes [1]

I was looking at this project a couple of months ago, and I did not understand why not use Kubernetes instead of rebuilding the abstractions. The reason is that Kubernetes already provides other abstractions to run services and batch job, gang scheduling, gpu and other accelerators enabled workflow.

[1]: https://polyaxon.com/docs/sandboxes/overview/

> a) I sound like an LLM when I'm writing articles (possible) or b) turing test AGI something something.

We entered the machinable culture. We spent many years trying to make the machine mimic humans, now humans are mimicking the machine :)

Actually msft CEO mentioned in his presentation that OAI moved to Azure's vector search and AI search services for ChatGPT.

I know that the use-case that I mentioned as well as many of the agentive aspects can be achieved using code. But I have to admit that using the UI and easily create GPTs, whether using them just as templates/personas or full-featured with actions/plugins, makes the use-case much easier, faster, and sharable. I can just @ at specific GPT to do something. Take the use-case that Simon mentions in his blog post, Dejargonizer, I can have a research GPT that helps with reviewin papers and I can @Dejargonizer to quickly explain a specific term, before resuming the discussion with the research GPT.

Maybe this would require additional research, but I think having a single GPT with access to all tools might be slower and less optimal, especially if the user knows exactly what they need for a given task and can reach for that quickly.

Yes, I had it pinned as soon as the UI changed post dev day.

One additional feature that I would like to see: interacting with 2 or more GPTs at the same time where they could perform different tasks based on their specific expertise and capabilities either in parallel or even sequentially as long as the replies/context of the discussion is accessible for further interactions, similar to what can be achieved with the assistants API.

> The default ChatGPT 4 UI has been updated: where previously you had to pick between GPT-4, Code Interpreter, Browse and DALL-E 3 modes, it now defaults to having access to all three. ... So I built Just GPT-4, which simply turns all three modes off, giving me a way to use ChatGPT that’s closer to the original experience.

Isn't that what they have already built-in called "ChatGPT classic". The description litteraly says "The latest version of GPT-4 with no additional capabilities"

Users can rate community notes.

Usually for asking questions about specific details, people are using RAG (Retrieval Augmented Generation) to ground the information and provide enough context for the llm to return the correct answers. This means additional engineering plumbing and very specific context to query information from.

You used the model for fact checking. These models are not good at being used as a knowledge base.

I see, thought you were using GPT-3.5 and moved to GPT-4.

> When OpenAI released GPT-4, we were able to change one parameter, and everything still just worked.

Wouldn't that be the same if you used the OAI js library directly? Basically swapping the model parameter?

Not recent, but the company that runs on top of milvus: https://www.businesswire.com/news/home/20220824005057/en/Vec...

Most demos shared by people can use numpy arrays similar to this https://twitter.com/karpathy/status/1647374645316968449

David Silver's reinforcement learning lecture series is excellent as well:

https://www.youtube.com/watch?v=2pWv7GOvuf0&list=PLzuuYNsE1E...

There is information on YC's website that answers your questions:

* https://www.ycombinator.com/about#yc-program-2

* https://www.ycombinator.com/faq

The open-source library is FastAPI. I might be wrong, but it's probably related to this tweet: https://twitter.com/tiangolo/status/1638683478245117953

They will probably have the full suite of Langchain features

I think the chat interface is a bit restrictive when it comes to multimodal models. A much cleaner interface would be an "AI notebook" where the user can move, compare, rerun blocks. Also sharing, versioning and collaborating with others on notebooks is more straightforward.

On your github org, twitter link is pointing to the wrong handle:

@oxen_ao -> @oxen_ai

A lot of teams use additional services from Github besides pushing code. They run their CI system using actions (including building docker images for prod), they keep knowledge/metadata on issues, PRs, comments, discussions, Kanban/project management, ...

Not having access to GH could be paralyzing.

How is Coder going to compete against GH/MSFT?

6 figures salaries for senior devs in Berlin are quite common, even since a couple years ago.

I would also recommend Khatia Buniatishvili's.

This tweet[1] has some insights on how Gen-Z users use TikTok for search.

[1]: https://twitter.com/AdriSheares/status/1557885461154111490

Streamlit is such a nice product. When iterating on an ML product and trying to give other users a quick and an easy way to interact with a model, Streamlit should be one of the first option to think of.

Integrating with Streamlit[1] was also very simple, in our case, we only had to expose how we serve Tensorboards and Notebooks on our platform, and we created a couple of tutorials[2] to show how to host an app. Several of our users started using it after that as the default way for sharing interactive and customizable dashboards on their Kubernetes clusters.

[1] https://polyaxon.com/integrations/streamlit/

[2] https://polyaxon.com/docs/intro/quick-start/deploying-ml-app...

I had some good success running similar but not as organized sessions for our open-source project[1] in 2018-2019. There was no specific day of the week and calls were between 30-50 min.

I think this format is much better and more organized, so I highly recommend that open-source maintainers who would like to learn more about how their projects are used to give it a try.

[1]: https://github.com/polyaxon/polyaxon

Probably OP was referring to the MPIOperator, TFOperator, PytorchOperator, ... they are under the Kuberflow org, but can be deployed independently of Kubeflow itself. Several other projects are using those operators to provide similar abstractions you mentioned in your blog post, e.g. Gang scheduling, cross-nodes communication, ...

One difference is that these operators use the Kubernetes service interface for communication, generally exposing a headless service for each replica.

Our tool provides several solutions, however, we do not force users to use all of these abstractions. It's very important for us that our product is interoperable with the rest of the ecosystem.

If a company is already using a pipelining tool, a visualization tool, or a data management tool, Polyaxon will work and integrate with those tools seamlessly.

That being said, and I fully understand where the OP is coming from, there are several companies not interested in managing several solutions and all the complexity that comes with the infrastructure, deployment, maintenance, upgrades, user facing clients, authn/authz, permissions... Polyaxon provides the right abstractions for covering the experimentation and the automation phase.