RulinShao / LightSeqLinks
Official repository for DistFlashAttn: Distributed Memory-efficient Attention for Long-context LLMs Training
☆221Updated last year
Alternatives and similar repositories for LightSeq
Users that are interested in LightSeq are comparing it to the libraries listed below
Sorting:
- Best practices for training DeepSeek, Mixtral, Qwen and other MoE models using Megatron Core.☆156Updated this week
- PyTorch bindings for CUTLASS grouped GEMM.☆140Updated 7 months ago
- Triton-based implementation of Sparse Mixture of Experts.☆262Updated 3 months ago
- PyTorch bindings for CUTLASS grouped GEMM.☆182Updated last month
- ☆115Updated last year
- GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM☆176Updated last year
- ☆89Updated 3 years ago
- [NeurIPS 2024] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization☆401Updated last year
- [ICLR 2025] COAT: Compressing Optimizer States and Activation for Memory-Efficient FP8 Training☆258Updated 5 months ago
- [MLSys'24] Atom: Low-bit Quantization for Efficient and Accurate LLM Serving☆335Updated last year
- Boosting 4-bit inference kernels with 2:4 Sparsity☆93Updated last year
- Bridge Megatron-Core to Hugging Face/Reinforcement Learning☆185Updated last week
- Ouroboros: Speculative Decoding with Large Model Enhanced Drafting (EMNLP 2024 main)☆113Updated 10 months ago
- ☆300Updated 6 months ago
- REST: Retrieval-Based Speculative Decoding, NAACL 2024☆215Updated 4 months ago
- Distributed IO-aware Attention algorithm☆23Updated 4 months ago
- [ICLR2025] Breaking Throughput-Latency Trade-off for Long Sequences with Speculative Decoding☆137Updated last year
- Zero Bubble Pipeline Parallelism☆448Updated 8 months ago
- Odysseus: Playground of LLM Sequence Parallelism☆79Updated last year
- [ICML 2024] Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference☆368Updated 6 months ago
- 🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.☆218Updated this week
- 16-fold memory access reduction with nearly no loss☆109Updated 10 months ago
- [NeurIPS 2024] The official implementation of "Kangaroo: Lossless Self-Speculative Decoding for Accelerating LLMs via Double Early Exitin…☆65Updated last year
- ☆353Updated last year
- An efficient GPU support for LLM inference with x-bit quantization (e.g. FP6,FP5).☆276Updated 6 months ago
- Accelerating MoE with IO and Tile-aware Optimizations☆553Updated last week
- nnScaler: Compiling DNN models for Parallel Training☆124Updated 4 months ago
- ☆158Updated 11 months ago
- Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity☆231Updated 2 years ago
- Code associated with the paper **Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding**☆216Updated 11 months ago