Next.js Hacker News
Show HN: Minimal LLM Post-Training Experiments on an 8GB GPU (SFT, DPO, GRPO)
15 points by popopanda 8 hours ago | discuss