Atlas · GenAI 2026
Hugging Face TRL
Transformer Reinforcement Learning library for SFT, DPO and RLHF/GRPO training.
toolPeak: 2024Fine-TuningAI consensus: 0/3
Prerequisites
- hardLLM Fine-Tuning
It is a fine-tuning/post-training library.
- mediumRLHF
It implements preference and RL training loops.
Recommended reference
TRL: Transformer Reinforcement Learning — Hugging Face documentation