Hugging Face Training Agents 3 explains GRPO for outcome-based agent training
Hugging Face · youtube · 2026-07-22
Hugging Face’s Training Agents 3 session covers reinforcement learning with GRPO, moving from a fine-tuned baseline to outcome-based training.
The session explains why GRPO comes after SFT and what RL adds beyond imitation. It walks through group sampling, relative advantages, and how GRPO replaces a critic or reward model by using the group’s relative scores as the learning signal. The agenda also includes writing verifiable reward functions for agent tasks, running GRPO in TRL on Hugging Face Jobs, and reading training curves such as reward, KL, and completion length.
The session also calls out common failure modes: reward hacking, collapse, and length gaming. It is part of a broader series about using coding agents to design, run, monitor, and review post-training experiments for better agentic models.
Related event: Hugging Face Live Stream Covers GRPO for Agent Reinforcement Learning(2 posts)→
More from coding & agent
- AI Automated Lead Gen: Targeting New Restaurants Before They Open — eptwts · 2026-07-23
- Open codebases are visible to everyone, but production still needs review — mmitchell_ai · 2026-07-23
- Browser automation stack pairs Browserbase, Stagehand, Weave evals, and Redis memory — wandb · 2026-07-23
- Parallel coding agents hit a human bottleneck: approvals, not compute — casualhermit · 2026-07-23
- Technical Debate: Should Base Transformers Come Equipped with Code Tools — a1zhang · 2026-07-23
- Webfetch claims 87% fewer search tokens and 66% lower cost for LLM agents — Remote-Breadfruit204 · 2026-07-23