Aligning Training and Rollout Policies for End-to-End FP4 RL on Mixture-of-Experts LLMs
cs.LG, cs.AI
Submitted: 2026-11-03
Updated: 2026-11-04
Terminology
Sources
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Jet-RL: Enabling On-Policy FP8 Reinforcement Learning with Unified Training and Rollout Precision Flow
- LoRA: Low-Rank Adaptation of Large Language Models
- HiFloat4 Format for End-To-End Reinforcement Learning Post-Training of Large Language Models
- DeepSeek-V3 Technical Report
- Mixtral of Experts
- Accurate INT8 Training Through Dynamic Block-Level Fallback
- Qwen3 Technical Report
- HybridFlow: A Flexible and Efficient RLHF Framework
- PyTorch FSDP: Experiences on Scaling Fully Sharded Data Parallel
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis