Show HN: Minimal LLM Post-Training Experiments on an 8GB GPU (SFT, DPO, GRPO)

 (github.com)

5 points | by popopanda 3 hours ago

1 comments