TRL, short for Transformers Reinforcement Learning, is a Hugging Face library for post-training transformer language models.
27
0
3,708 live tools · ranked by monthly traffic and community votes · dead links removed automatically