The sad part is that people who know what they do not know are usually in the academia... People who do not know what they do not know are usually the loudest and the least serious...
The sad part is that people who know what they do not know are usually in the academia... People who do not know what they do not know are usually the loudest and the least serious...
(1/8)🍎A Galileo moment for LLM design🍎
As Pisa Tower experiment sparked modern physics, our controlled synthetic pretraining playground reveals LLM architectures' true limits. A turning point that might divide LLM research into "before" and "after." physics.allen-zhu.com/part-4-archite…
LIMO: Less is More for Reasoning
Achieves 57.1% on AIME and 94.8% on MATH w/ only 817 training samples, i.e., only 1% of the training data required by previous approaches
Chain-of-Associated-Thoughts (CoAT) is a new framework that enhances LLMs' reasoning abilities by combining Monte Carlo Tree Search with dynamic knowledge integration.
The framework addresses the limitations of existing "fast thinking" approaches by introducing an "associative…
Excited to introduce flow Q-learning (FQL)!
Flow Q-learning is a *simple* and scalable data-driven RL method that trains an expressive policy with flow matching.
Paper: arxiv.org/abs/2502.02538
Project page: seohong.me/projects/fql/
Thread ↓
2K Followers 639 FollowingFounder @Unum_Cloud → building fastest open-source AI infra one Assembly instruction at a time • Logs: https://t.co/qbiT4hUzoL • Investments: https://t.co/bSGttd0qGP
49K Followers 224 FollowingEx HTX/Huobi CFO, Ex OKX CEO n Gp CFO| Web3 AI Fund | Best CFO II| Best HK Exe Asiamoney| ‘20 Buffett Charity Dinner| Opinions Are My Own.
2K Followers 22 FollowingRAGFlow is the leading open-source RAG engine, converging cutting-edge RAG with Agent to build context layer for LLMs.
Discord: https://t.co/VqH4a1qqPE
162K Followers 461 FollowingTwitter is my Chain-Of-Thought. Reading history is my end-to-end training. Not financial advice. 一言不合就拉黑。评论区只对订阅用户开放。
Runner: 1 km, 3'49; 5 km, 23'07
365K Followers 6K FollowingChief Scientist, Google DeepMind & Google Research. Gemini Lead. Opinions stated here are my own, not those of Google. TensorFlow, MapReduce, Bigtable, ...
163K Followers 0 FollowingInvented principles of meta-learning (1987), GANs (1990), Transformers (1991), very deep learning (1991), etc. Our AI is used many billions of times every day.
25K Followers 101 FollowingDirector, @PrincetonPLI and Professor @PrincetonCS. Seeks math/conceptual understanding of deep learning and large AI models.
Also on the "other" social network
11K Followers 1K FollowingMarketer, self-taught developer, and founder of @Bento and Tatami. Designing a quiet family life in 福岡, Japan. DMs open if you need email help 🌿
13K Followers 688 FollowingResearch @Meta Superintelligence Labs, RL/post-training/agents; Previously Research @OpenAI on multimodal and RL; Opinions are my own.
37K Followers 565 FollowingAssistant professor at Stanford; Co-founder of Voyage AI (https://t.co/wpIITHLgF0) ;
Working on ML, DL, RL, LLMs, and their theory.
8K Followers 451 FollowingProfessor @MITEECS and @MIT_CSAIL. Computational complexity, algorithm design, and related math. I'll let you know when P != NP is proved (and when it's not)
9K Followers 2K FollowingMachine learning scientist and engineer speaking πtorch & C++. Past @LTIatCMU, @awscloud. Opinions sampled from MY OWN 100T param LM.