Akram Baharlouei

ai/ml @altos_labs | x @Meta AI & Qualcomm

[Akram Baharlouei @yalda2009]

Great discussion on self supervised learning with awesome people! Thank you all for attending. I enjoyed and learned a lot.🙏🙏 really impressed by more than 400 people attending on such a special topic:)

[Akram Baharlouei @yalda2009]

LLM sessions @ #NeurIPS23

[Akram Baharlouei @yalda2009]

@ #NeurIPS2023 , who’s around? Happy to meet old and new friends 👋

[Akram Baharlouei @yalda2009]

@jerryjliu0 @disiok @Forbes @llama_index Congrats!!!

[Akram Baharlouei @yalda2009]

@ShunyuYao12 Q-learning + A* I suppose?

[Jerry Liu @jerryjliu0]

The @OpenAI retrieval API seems to be doing basic top-k RAG on limited context if there's context overflows.

Asking for a summary of the Uber 10-K (~150k-200k tokens), returns oddly specific stuff around acquisitions and legal proceedings 👇

[Akram Baharlouei @yalda2009]

@ahkashani @ServeRobotics @Uber @nvidia Congrats Ali!

[Andrej Karpathy @karpathy]

Who’s getting how many H100s and when is top gossip of the valley rn

[Andrej Karpathy @karpathy]

Huge day indeed for AI and LLMs, congrats to Meta 👏
This is now the most capable LLM available directly as weights to anyone from researchers to companies.

The models look quite strong, e.g. Table 4 in the paper: MMLU is good to look at, the 70B model is just below GPT-3.5. But HumanEval (bad misnomer) shows coding capability is quite a bit lower (48.1 vs 29.9).

[Akram Baharlouei @yalda2009]

@ Data+AI Summit conference. Who’s around? Let’s connect! #DataAISummit #Mosconecenter

[Akram Baharlouei @yalda2009]

@Data_AI_Summit LLMs, gen AI in production!

[Akram Baharlouei @yalda2009]

@Data_AI_Summit conference. Who’s around? Let’s connect! #DataAISummit

[Justin Johnson @jcjohnss]

I'm excited about Segment Anything released from FAIR today. It tackles an old problem (find objects in images) at large scale: trained on 11M images and 1B objects.
This is a new Foundation Model for Computer Vision - it recognizes any object in any context.

[Jeff Dean @JeffDean]

Excited to share the first of a series of @GoogleAI blog posts summarizing our research work from 2022. This covers language & multimodal models, computer vision, and generative models. We'll have ~7 posts covering other areas over next few weeks!

https://t.co/WuHSZ6qKKu

[Akram Baharlouei @yalda2009]

@krandiash Lol

[Akram Baharlouei @yalda2009]

@mariyaivasileva Third

[Akram Baharlouei @yalda2009]

Happening now!!!

[AI at Meta @AIatMeta]

And we're live! Tune in as we explore the future of #AI: https://t.co/UsMV4XtSgQ

[AI at Meta @AIatMeta]

And we're live! Tune in as we explore the future of #AI: https://t.co/UsMV4XtSgQ

[Oriol Vinyals @OriolVinyalsML]

This is a great thread, +1e100. @karpathy didn't mention my (biased!) favorite example of 2021. Models designed to generate words (transformers) & model language (BERT) were reused in #AlphaFold to solve the protein folding problem, mapping a bunch of letters, to 3D coordinates🤯

[Yann LeCun @ylecun]

Which is why the best preparation for a PhD in ML/AI is not necessarily CS but electrical engineering, physics, or applied mathematics.

[Akram Baharlouei @yalda2009]

@gdb Wow!!!