# FP8-RL: A Practical and Stable Low-Precision Stack for LLM Reinforcement Learning

arXiv: https://arxiv.org/abs/2601.18150 · HTML: https://mlsanity.com/paper/2601.18150/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Placed in Speculative decoding, one of 12 directions in Efficient LLM inference and serving.
- Builds on 10 of its 21 references that are in the map.
- First posted on arXiv 26 Jan 2026.

## Placed in

- Efficient LLM inference and serving / [Speculative decoding](https://mlsanity.com/inference/speculative-decoding/)

## Builds on (most cited first, 10 of 10)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://mlsanity.com/paper/2409.19256/) (2024)
- [AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning](https://mlsanity.com/paper/2505.24298/) (2025)
- [FP8 Formats for Deep Learning](https://mlsanity.com/paper/2209.05433/) (2022)
- [DeepSpeed-Chat: Easy, Fast and Affordable RLHF Training of ChatGPT-like Models at All Scales](https://mlsanity.com/paper/2308.01320/) (2023)
- [Reinforcement Learning Optimization for Large-Scale Learning: An Efficient and User-Friendly Scaling Library](https://mlsanity.com/paper/2506.06122/) (2025)
- [FP8-LM: Training FP8 Large Language Models](https://mlsanity.com/paper/2310.18313/) (2023)
- [Stabilizing MoE Reinforcement Learning by Aligning Training and Inference Routers](https://mlsanity.com/paper/2510.11370/) (2025)
- [Seer: Online Context Learning for Fast Synchronous LLM Reinforcement Learning](https://mlsanity.com/paper/2511.14617/) (2025)

## Built on by (most cited first, 13 of 13)

- [Accelerating RL Post-Training Rollouts via System-Integrated Speculative Decoding](https://mlsanity.com/paper/2604.26779/) (2026)
- [Diagnosing Training Inference Mismatch in LLM Reinforcement Learning via a Zero-Mismatch Reference](https://mlsanity.com/paper/2605.14220/) (2026)
- [FP4 Explore, BF16 Train: Diffusion Reinforcement Learning via Efficient Rollout Scaling](https://mlsanity.com/paper/2604.06916/) (2026)
- [AIS: Adaptive Importance Sampling for Quantized RL](https://mlsanity.com/paper/2605.13907/) (2026)
- [QUADS: Stabilizing NVFP4 Reinforcement Learning for MoE via QUantization-error Alignment across Dual Sides](https://mlsanity.com/paper/2607.15810/) (2026)
- [Score Centering Stabilizes Off-policy Reinforcement Learning](https://mlsanity.com/paper/2609.20807/) (2026)
- [LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction](https://mlsanity.com/paper/2608.05600/) (2026)
- [Towards Full Pipeline FP8 Reinforcement Learning for LLMs](https://mlsanity.com/paper/2609.22870/) (2026)
- [Rollout Efficiency in Reinforcement Learning for Reasoning Large Language Models: A Taxonomy and Future Directions](https://mlsanity.com/paper/2609.25463/) (2026)
- [HiLoRe: What to Store, Compress, or Recompute for Efficient GRPO Training](https://mlsanity.com/paper/2609.33570/) (2026)
- [TRIAGE: Direction-Aware Mismatch Stabilization of Native NVFP4 Reinforcement Learning](https://mlsanity.com/paper/2610.07043/) (2026)
- [KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization](https://mlsanity.com/paper/2610.14448/) (2026)
- [ObsSpec: Accelerating Agentic Reinforcement Learning with Observation Speculation](https://mlsanity.com/paper/2610.15246/) (2026)

## Abstract (arXiv)

Reinforcement learning (RL) for large language models (LLMs) is increasingly bottlenecked by rollout (generation), where long output sequence lengths make attention and KV-cache memory dominate end-to-end step time. FP8 offers an attractive lever for accelerating RL by reducing compute cost and memory traffic during rollout, but applying FP8 in RL introduces unique engineering and algorithmic challenges: policy weights change every step (requiring repeated quantization and weight synchronization into the inference engine) and low-precision rollouts can deviate from the higher-precision policy assumed by the trainer, causing train-inference mismatch and potential instability. This report presents a practical FP8 rollout stack for LLM RL, implemented in the veRL ecosystem with support for common training backends (e.g., FSDP/Megatron-LM) and inference engines (e.g., vLLM/SGLang). We (i) enable FP8 W8A8 linear-layer rollout using blockwise FP8 quantization, (ii) extend FP8 to KV-cache to remove long-context memory bottlenecks via per-step QKV scale recalibration, and (iii) mitigate mismatch using importance-sampling-based rollout correction (token-level TIS/MIS variants). Across dense and MoE models, these techniques deliver up to 44% rollout throughput gains while preserving learning behavior comparable to BF16 baselines.