yaof20 / Flash-RLLinks
Implementation for FP8/INT8 Rollout for RL training without performence drop.
☆288Updated 2 months ago
Alternatives and similar repositories for Flash-RL
Users that are interested in Flash-RL are comparing it to the libraries listed below
Sorting:
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference☆161Updated 3 months ago
- Bridge Megatron-Core to Hugging Face/Reinforcement Learning☆188Updated this week
- ☆129Updated 7 months ago
- ☆220Updated 2 months ago
- SeerAttention: Learning Intrinsic Sparse Attention in Your LLMs☆187Updated 4 months ago
- Training library for Megatron-based models with bidirectional Hugging Face conversion capability☆386Updated last week
- Async pipelined version of Verl☆124Updated 9 months ago
- [ASPLOS'26] Taming the Long-Tail: Efficient Reasoning RL Training with Adaptive Drafter☆130Updated last month
- [ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training☆258Updated 5 months ago
- ☆133Updated 8 months ago
- 🔥 A minimal training framework for scaling FLA models☆341Updated 2 months ago
- [ICLR 2025] PEARL: Parallel Speculative Decoding with Adaptive Draft Length☆148Updated last month
- Spectral Sphere Optimizer☆90Updated 2 weeks ago
- A lightweight reinforcement learning framework that integrates seamlessly into your codebase, empowering developers to focus on algorithm…☆98Updated 5 months ago
- 🔥 LLM-powered GPU kernel synthesis: Train models to convert PyTorch ops into optimized Triton kernels via SFT+RL. Multi-turn compilation…☆115Updated 2 months ago
- Efficient triton implementation of Native Sparse Attention.☆261Updated 8 months ago
- Official PyTorch implementation of the paper "dLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Caching" (dLLM-Cache…☆197Updated 2 months ago
- APRIL: Active Partial Rollouts in Reinforcement Learning to Tame Long-tail Generation. A system-level optimization for scalable LLM tra…☆47Updated 3 months ago
- ☆110Updated 4 months ago
- 16-fold memory access reduction with nearly no loss☆109Updated 10 months ago
- [ICLR2025] Codebase for "ReMoE: Fully Differentiable Mixture-of-Experts with ReLU Routing", built on Megatron-LM.☆104Updated last year
- Super-Efficient RLHF Training of LLMs with Parameter Reallocation☆330Updated 9 months ago
- qwen-nsa☆87Updated 3 months ago
- ☆300Updated 6 months ago
- PoC for "SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning" [NeurIPS '25]☆61Updated 3 months ago
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆368Updated 6 months ago
- ☆104Updated 11 months ago
- LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification☆73Updated 6 months ago
- ByteCheckpoint: An Unified Checkpointing Library for LFMs☆264Updated last month
- ☆118Updated 4 months ago