1 citations · 1 across the 6 of their papers we have counts for
4 papers · 1 filter
Single-stream Policy Optimization
Zhongwen Xu, Zihan Ding
We revisit policy-gradient optimization for Large Language Models (LLMs) from a single-stream perspective. Prevailing group-based methods like GRPO reduce variance with on-the-fly…
Understanding Tool-Integrated Reasoning
Heng Lin, Zhongwen Xu
We study why Tool-Integrated Reasoning (TIR) makes Large Language Models (LLMs) more capable. While LLMs integrated with tools like Python code interpreters show great promise, a p…
Agents Play Thousands of 3D Video Games
Zhongwen Xu, Xianliang Wang, Siyi Li +4
We present PORTAL, a novel framework for developing artificial intelligence agents capable of playing thousands of 3D video games through language-guided policy generation. By tran…
Cleanba: A Reproducible and Efficient Distributed Reinforcement Learning Platform
Shengyi Huang, Jiayi Weng, Rujikorn Charakorn +3
Distributed Deep Reinforcement Learning (DRL) aims to leverage more computational resources to train autonomous agents with less training time. Despite recent progress in the field…