Training RL/robot policies requires extensive experience in the target environment, which is often difficult to obtain. How can we “distill” embodied policies from foundational models? Introducing FactorSim! #NeurIPS2024 We show that by generating prompt-aligned simulations and

if you are coming to cscw: our talk is next tuesday at 11 AM in session 2c! mark your calendar!

hey! i’m at #cscw2024 presenting our paper on tuesday at 11am! we built a browser extension helping users avoid dark patterns on their interfaces! it’s cool! you should check it out!! it also got a best paper award! picture is today @ la paz waterfall garden, 🇨🇷 is amazing

✨EMNLP Paper ✨ Wouldn't it be great if we can also listen to LLM responses when we can't look at a screen? Problem: LLMs generate responses without considering the unique constraints of speech 😢 🎉 Let's fix that with Speechworthy Instruction-tuned Language Models

I'm presenting this work during today's poster session from 10:30AM-12PM at EMNLP! Come by and say hi 👋 x.com/HJCH0/status/1…

As many of you might already know, I had this dream to become a faculty. Unfortunately, that did not happen, so instead I decided to create my own lab kind of a thing. I am trying to start working with some students (PhD, masters, undergrad are all welcome) and mentor them.

Check us out at NeurIPS tomorrow! Unfortunately I can’t be there but Logan Cross and Jonathan will present at East Exhibit Hall A-C

I will deliver the final talk in the SPS SLTC/AASP TC Webinar Series in 2024, sharing insights on fine-tuning models such as LLaMA and Whisper. 📅 Dec 17, 10 AM ET (11 PM Taiwan). Register: signalprocessingsociety.org/blog/sps-sltca…

check out our new work on merging expert models, Branch-Train-Stitch 🪡🪡🪡 had so much fun working on this with the incredible Qizhen (Irene) Zhang and team!!! 😊

(1/5)🚨LLMs can now self-improve to generate better citations✅ 📝We design automatic rewards to assess citation quality 🤖Enable BoN/SimPO w/o external supervision 📈Perform close to “Claude Citations” API w/ only 8B model 📄arxiv.org/abs/2502.09604 🧑‍💻github.com/voidism/SelfCi…

🚀Happy to share my last-year  Apple's internship work! A promising use case of long-context LLMs is enabling the entire knowledge base to fit in the prompt as contextual knowledge for tasks like QA, rather than RAG pipeline. But are they up to this? If not, how to improve?

🚀 New Paper Alert! 🚀 Want better LLM-as-a-Judge? TRACT: 🧠 CoT + Regression-Aware Fine-tuning (RAFT) = Better numerical predictions! 📊 arxiv.org/abs/2503.04381 🧵👇 A thread on TRACT:

Spatial reasoning is a major challenge for the foundation models today, even in simple tasks like arranging objects in 3D space. #CVPR2025 Introducing LayoutVLM, a differentiable optimization framework that uses VLM to spatially reason about diverse scene layouts from unlabeled

Scaling Laws of Synthetic Data for Language Models SynthLLM is a framework that generates high-quality synthetic pretraining data by extracting and recombining concepts across documents using a graph algorithm. It follows predictable scaling laws, with performance gains

We invite in-person tutorial proposals to the Asian Conference on Machine Learning (ACML) 2025 in Taipei, Taiwan, on Dec 12, 2025! Share your research with us & visit vibrant Taiwan! #ACML2025 Deadline: Aug 1; notification: Sep 5 CFT: acml-conf.org/2025/tutorial.… Please retweet!

🎧 With the rapid growth of audio LLM benchmarking studies, a comprehensive survey is timely! Check out the survey paper on benchmarks in audio LLMs by Chih-Kai Yang and Neo S. Ho. 🔥 Paper link: arxiv.org/abs/2505.15957