Reproduce “Agentic Critical Training” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Weize Liu, Minghui Liu, Sy-Tuyen Ho, Yongkyun Lee, Andrew Adams Schoen, Souradip Chakraborty, Xiyao Wang, Furong Huang
ACT trains agents to compare an expert action with an alternative sampled from their initial policy, then uses imitation learning and optionally reinforcement learning. It improves results across three benchmarks and two model families by mean gains of 5.85 points over imitation learning and 4.12 over imitation learning followed by reinforcement learning. These are average point changes across different metrics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.