Many might not know. John Schulman is the main person driving ChatGPT, and I would like to congratulate him and the team! Worth to follow.

Potentially nitpicky but competitive advantage in AI goes not so much to those with data but those with a data engine: iterated data aquisition, re-training, evaluation, deployment, telemetry. And whoever can spin it fastest. Slide from Tesla to ~illustrate but concept is general

RL from human feedback seems to be the main tool for alignment. Given reward hacking and the falliability of humans, this strategy seems bound to produce agents that merely appear to be aligned, but are bad/wrong in subtle, inconspicuous ways. Is anyone else worried about this?

Many believe that great AI advances must contain a new “idea”. But it is not so: many of AI’s greatest advances had the form “huh, turns out this familiar unimportant idea, when done right, is downright incredible”

The entire prompt of Microsoft Bing Chat?! (Hi, Sydney.)

New open source Flan-UL2 20B checkpoints :) - Truly open source 😎 No forms! 🤭 Apache license 🔥 - Best OS model on MMLU/Big-Bench hard 🤩 - Better than Flan-T5 XXL & competitive to Flan-PaLM 62B. - Size ceiling of Flan family just got higher! Blog: yitay.net/blog/flan-ul2-…

In the Transformer movies, 9 Decepticons merge to form “Devastator”, a much larger and stronger bot. This turns out to be a powerful paradigm for multimodal LLM too. Instead of a monolithic Transformer, we can stack many pre-trained experts into one. My team’s work, Prismer, is

I believe next-gen LLMs will heavily borrow insights from a decade of game AI research. ▸ Noam Brown, creator of Libratus poker AI, is joining OpenAI. ▸ Demis Hassabis says that DeepMind Gemini will tap techniques from AlphaGo. These moves make a lot of sense. Methods like

# RLHF is just barely RL Reinforcement Learning from Human Feedback (RLHF) is the third (and last) major stage of training an LLM, after pretraining and supervised finetuning (SFT). My rant on RLHF is that it is just barely RL, in a way that I think is not too widely

Great keynote by David Silver, arguing that we need to re-focus on RL to get out of the LLM Valley RL_Conference

Mini-Omni, an open-source real-time audio conversational model ⚡️Real-time conversational speech-to-speech 🤯Can generate text and audio at the same time 🚀Streaming audio output Model: hf.co/gpt-omni/mini-… Paper: hf.co/papers/2408.16… Codebase: github.com/gpt-omni/mini-…

I think LLM offers an incredible opportunity for the next generation of RL works. The LLM is like the imitation learning policy in AlphaStar, and we need to do some cool RL stuff to make more magic. Lots of exciting work ahead!

Super-excited about what's ahead. Want to move the AI research frontier, join us! amazon.science/blog/amazon-op… [email protected]

"Move 37" is the word-of-day - it's when an AI, trained via the trial-and-error process of reinforcement learning, discovers actions that are new, surprising, and secretly brilliant even to expert humans. It is a magical, just slightly unnerving, emergent phenomenon only

For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.

Excited to share a research preview of what the team’s building: a browser-using AI agent for reliability and usefulness. Check it out and join us on this journey → labs.amazon.science

Stoked about the first release from our new lab: our browser use agent lets you MapReduce over the web! This early preview moves us closer to reliable agents that learn from rewards across a wide range of digital and physical environments. Love our Adept+Amazon team so much!

I'm thrilled to share our first release as the AGI SF Lab. Meet Nova Act -- the most effortless way to build agents that can reliably use browsers, giving agents access to much of our digital world. It brings us closer to building universal agents in both digital and physical

There are traditionally two types of research: problem-driven research and method-driven research. As we’ve seen with large language models and now AlphaEvolve, it should be very clear now that total method-driven research is a huge opportunity. Problem-driven research is nice

It’s funny that people on this site think major LLM efforts are talent-bound rather than org-bound. The talent differential has never been big between major orgs. Most of the difference in outcomes is due to organisational factors - like allocating compute to the right bets, and