li-plus / flash-preferenceLinks
Accelerate LLM preference tuning via prefix sharing with a single line of code
☆46Updated 3 months ago
Alternatives and similar repositories for flash-preference
Users that are interested in flash-preference are comparing it to the libraries listed below
Sorting:
- Odysseus: Playground of LLM Sequence Parallelism☆78Updated last year
- Quantized Attention on GPU☆44Updated 11 months ago
- ☆130Updated 4 months ago
- ☆50Updated 5 months ago
- ☆101Updated 5 months ago
- An efficient implementation of the NSA (Native Sparse Attention) kernel☆121Updated 4 months ago
- 🔥 LLM-powered GPU kernel synthesis: Train models to convert PyTorch ops into optimized Triton kernels via SFT+RL. Multi-turn compilation…☆86Updated 2 weeks ago
- A lightweight reinforcement learning framework that integrates seamlessly into your codebase, empowering developers to focus on algorithm…☆68Updated 2 months ago
- [ACL 2024] RelayAttention for Efficient Large Language Model Serving with Long System Prompts☆40Updated last year
- A Suite for Parallel Inference of Diffusion Transformers (DiTs) on multi-GPU Clusters☆50Updated last year
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆130Updated 10 months ago
- The evaluation framework for training-free sparse attention in LLMs☆102Updated 2 weeks ago
- Beyond KV Caching: Shared Attention for Efficient LLMs☆19Updated last year
- Flash-Muon: An Efficient Implementation of Muon Optimizer☆197Updated 4 months ago
- Xmixers: A collection of SOTA efficient token/channel mixers☆29Updated last month
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference☆147Updated 2 weeks ago
- 16-fold memory access reduction with nearly no loss☆105Updated 7 months ago
- ☆39Updated 2 months ago
- ☆120Updated 2 months ago
- ☆25Updated 6 months ago
- ☆145Updated 8 months ago
- Estimate MFU for DeepSeekV3☆26Updated 9 months ago
- [CoLM'25] The official implementation of the paper <MoA: Mixture of Sparse Attention for Automatic Large Language Model Compression>☆147Updated 3 months ago
- The official implementation of paper: SimLayerKV: A Simple Framework for Layer-Level KV Cache Reduction.☆49Updated last year
- LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification☆65Updated 3 months ago
- Framework to reduce autotune overhead to zero for well known deployments.☆84Updated last month
- ☆65Updated 6 months ago
- Transformers components but in Triton☆34Updated 5 months ago
- Kinetics: Rethinking Test-Time Scaling Laws☆81Updated 3 months ago
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆169Updated last year