Orbnet was the first physics-informed ML for quantum chemistry that can extrapolate to molecules significantly larger than the ones in training data
@animaanandkumar
-
Frances Arnold Inspires Innovation in AI Research
By
–
What an amazing story! @francesarnold continues to inspire me everyday. A must read!
-
LeanAgent: Lifelong Learning Agent for Formal Theorem Proving
By
–
We are excited to share our #ICLR2025 paper on LeanAgent: the first lifelong learning agent for formal theorem proving in Lean. LLMs have been integrated with interactive proof assistants like Lean for theorem proving with 100% accuracy. So far, these LLMs cannot continuously
-
HeadInfer: Long-Context LLM Inference on Consumer GPUs
By
–
HeadInfer: Unlocking Long-Context LLM Inference on Consumer GPUs (Million-level Tokens)
*long-context inputs require large GPU memory.
*A standard LLM like Llama-3–8B requires 207GB of GPU memory for 1 million tokens — far beyond the capabilities of consumer GPUs like the RTX -
LLMs Breakthrough: LeanProgress Critic Model Enhances Formal Proof Search
By
–
LLMs still struggle with long theorem proofs. LeanProgress offers a solution. We introduce a critic model for Lean4 proof search, and it uses a distance-based measure. Instead of just log-probability (logp), our critic uses “distance” as a signal, guiding the LLM to explore and
-
TIME 100 Impact Award Recognition in AI Community
By
–
It was a pleasant surprise to be receiving the @TIME 100 impact award. I am in great company with @ArvindKrishna @Grimezsz @refikanadol and prior awardees @ylecun @KayFButterfield @PadmaLakshmi @AshleyJudd @ayushmannk @aliaa08 @deepikapadukone and many other fabulous people
-
TIME 100 Impact Award Recognition for AI Science Work
By
–
What a special evening at @TIME 100 Impact Award dinner last night. Giving the acceptance speech https://
youtu.be/uX3hC9xJqBQ was really touching, especially with my parents in the audience. Thank you @TIME for shining the spotlight on my work in AI+Science. This is just the -
Language Models Lack Physical Grounding for Scientific Discovery
By
–
Text knowledge is not sufficient for scientific discovery. Language models lack physical grounding. They only have a high level of understanding but cannot simulate physical phenomena. Image and video models focus on "looking good" vs. being physically valid. We’re teaching
-
Robust Representation Consistency Model Accepted to ICLR 2025
By
–
Thrilled to announce our paper "Robust Representation Consistency Model (rRCM)" was accepted to #ICLR2025.
It combines contrastive learning with consistency training to enhance robust representation learning and sets a new standard in certified robustness. @JiachenLei @julberner -
Neural Operators and AI Science Recognized by TIME
By
–
Thrilled to receive this honor by @TIME Great to see neural operators and AI + science be recognized!