fp8

Towards Full Pipeline FP8 Reinforcement Learning for LLMsarxiv.org

FP8 quantization of RL pipelines for LLMs causes training instability because quantization noise skews importance ratios, zeroing gradients for negative‑advantage tokens and leading to entropy spikes. The authors introduce Calibrated Clipping, a dynamic adjustment of FP8 clipping bounds that matches BF16 quantile statistics, restoring stable training across GRPO and DAPO methods for 8‑32 B models…