eto-sft-trajectory

Expert Trajectories for ETO 🌐 Homepage | 🐍 GitHub | 📖 arXiv Expert trajectories for Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents Authors: Yifan Song, Da Yin, Xiang Yue, Jie Huang, Sujian Li, Bill Yuchen Lin. We introduce ETO (Exploration-based Trajectory Optimization), an agent learning framework inspired by "trial and error" process of human learning. ETO allows an LLM agent to iteratively collect failure trajectories and updates its policy by… See the full description on the dataset page: https://huggingface.co/datasets/agent-eto/eto-sft-trajectory.