RESEARCH - AI Dynamics

BBF: Value-Based RL Agent Achieves Super-Human Atari Performance

By

–

01 June 2023 6h02

Bigger, Better, Faster: Human-level Atari with human-level efficiency introduce a value-based RL agent, which we call BBF, that achieves super-human performance in the Atari 100K benchmark. BBF relies on scaling the neural networks used for value estimation, as well as a number

→ View original post on X — @_akhaliq

1 June 2023

Improving CLIP Training with Language Rewrites via LaCLIP

By

@_akhaliq

–

01 June 2023 5h18

Improving CLIP Training with Language Rewrites introduce Language augmented CLIP (LaCLIP), a simple yet highly effective approach to enhance CLIP training through language rewrites. Leveraging the in-context learning capability of large language models, we rewrite the text

→ View original post on X — @_akhaliq

1 June 2023

Open LLM Leaderboard on Hugging Face

By

@abhi1thakur

–

01 June 2023 5h08

And the Open LLM LB: https://
huggingface.co/spaces/Hugging
FaceH4/open_llm_leaderboard
… 3/

→ View original post on X — @abhi1thakur

1 June 2023

Falcon 40B Model Now Available on Hugging Face

By

@abhi1thakur

–

01 June 2023 5h08

Check out Falcon 40B model here: https://
huggingface.co/tiiuae/falcon-
40b
… 2/

→ View original post on X — @abhi1thakur

1 June 2023

Falcon LLM Goes Open-Source with Apache 2.0 License

By

@abhi1thakur

–

01 June 2023 5h08

From 10% royalty to 0%! Falcon (40B and 7B) goes truly open-source with Apache 2.0 license. Falcon 40B currently leads the Open LLM leaderboard beating Facebook's LLAMA 65B model. Open-source will rule! P.S. Image was generated using SD2.1 😉 1/

→ View original post on X — @abhi1thakur

1 June 2023

Process Supervision: Training AI Through Correct Reasoning Steps

By

@gdb

–

01 June 2023 0h55

Initial results with process supervision — training an AI by rewarding it for coming up with a correct thought process, not just outputting the right answer. Great results in mathematical reasoning:

→ View original post on X — @gdb

1 June 2023

OpenAI Improves Math Reasoning with RLHF and Chain-of-Thought

By

@alexandr_wang

–

01 June 2023 0h31

research from @OpenAI on improving math reasoning by RLHF with a reward model trained on 800k human-generated chain-of-thought data (which @scale_AI partnered w/
@OpenAI on!) RLHF seems to be a scalable technique for making LLMs smarter in many ways

→ View original post on X — @alexandr_wang

1 June 2023

How AI Can Transform The Software Engineering Process

By

@terence_mills

–

01 June 2023 0h30

How AI Can Transform The Software Engineering Process
#AI #AIio #BigData #ML #NLU #Futureofwork @briansolis @Xbond49 @VitalikButerin @DataScienceCtrl @Benioff @johnnosta @IrmaRaste @Tiffani_Bova http://
ow.ly/U09430svtq3

→ View original post on X — @terence_mills

1 June 2023

Mark Zuckerberg Podcast Interview: AI and Beyond

By

@lexfridman

–

31 May 2023 23h40

I'm interviewing Mark Zuckerberg on the podcast soon, for the 2nd time. The focus will be on AI , but I'm sure we'll cover a lot of other topics as well. Please let me know if you have any questions/topic suggestions.

→ View original post on X — @lexfridman

31 May 2023