DeepSeek Native Sparse Attention pytorch implementation
☆114Dec 17, 2025Updated 2 months ago
Alternatives and similar repositories for NSA-pytorch
Users that are interested in NSA-pytorch are comparing it to the libraries listed below
Sorting:
- 🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"☆969Feb 5, 2026Updated 3 weeks ago
- qwen-nsa☆87Oct 14, 2025Updated 4 months ago
- An efficient implementation of the NSA (Native Sparse Attention) kernel☆129Jun 24, 2025Updated 8 months ago
- Sample Codes using NVSHMEM on Multi-GPU☆30Jan 22, 2023Updated 3 years ago
- ☆111Updated this week
- LongAttn ：Selecting Long-context Training Data via Token-level Attention☆15Jul 16, 2025Updated 7 months ago
- Expert Specialization MoE Solution based on CUTLASS☆27Jan 19, 2026Updated last month
- Complete simulation of IEEE 754 fixed and floating point specification to any precision☆12Aug 26, 2020Updated 5 years ago
- Website for CSE 234, Winter 2025☆13Mar 24, 2025Updated 11 months ago
- TileGraph is an experimental DNN compiler that utilizes static code generation and kernel fusion techniques.☆12Sep 18, 2024Updated last year
- ☆68Jun 23, 2025Updated 8 months ago
- CUDA SGEMM optimization note☆15Oct 31, 2023Updated 2 years ago
- PyTorch implementation of the Flash Spectral Transform Unit.☆21Sep 19, 2024Updated last year
- [ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads☆524Feb 10, 2025Updated last year
- Wave: Python Domain-Specific Language for High Performance Machine Learning☆45Updated this week
- Algorithms for approximate attention in LLMs☆21Apr 14, 2025Updated 10 months ago
- DeepSeek-V3.2-Exp DSA Warmup Lightning Indexer training operator based on tilelang☆44Nov 19, 2025Updated 3 months ago
- ☆44Updated this week
- ☆39Dec 14, 2025Updated 2 months ago
- ☆19Sep 19, 2024Updated last year
- Creating the DeepSeek V3 model from scratch☆25Mar 28, 2025Updated 11 months ago
- ☆49Feb 5, 2026Updated 3 weeks ago
- This repository is a collection of legal instruction datasets☆26Jul 12, 2024Updated last year
- 最简易的R1结果在小模型上的复现，阐 述类O1与DeepSeek R1最重要的本质。Think is all your need。利用实验佐证，对于强推理能力，think思考过程性内容是AGI/ASI的核心。☆45Feb 8, 2025Updated last year
- Tile-Based Runtime for Ultra-Low-Latency LLM Inference☆663Feb 14, 2026Updated 2 weeks ago
- TransMLA: Multi-Head Latent Attention Is All You Need (NeurIPS 2025 Spotlight)☆434Updated this week
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.☆106Jun 28, 2025Updated 8 months ago
- Awesome code, projects, books, etc. related to CUDA☆31Feb 3, 2026Updated last month
- Risk Management via Anomaly Circumvent: Mnemonic Deep Learning for Midterm Stock Prediction. KDD 2019.☆23Aug 26, 2020Updated 5 years ago
- A collection of memory efficient attention operators implemented in the Triton language.☆288Jun 5, 2024Updated last year
- ☆134May 29, 2025Updated 9 months ago
- Efficient LLM Inference over Long Sequences☆393Jun 25, 2025Updated 8 months ago
- [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.☆952Updated this week
- Distributed IO-aware Attention algorithm☆24Sep 24, 2025Updated 5 months ago
- ☆118May 19, 2025Updated 9 months ago
- Supplemental materials for The ASPLOS 2025 / EuroSys 2025 Contest on Intra-Operator Parallelism for Distributed Deep Learning☆25May 12, 2025Updated 9 months ago
- ☆272Jun 6, 2025Updated 8 months ago
- [AAAI 2026 Oral] The official code of "UniME-V2: MLLM-as-a-Judge for Universal Multimodal Embedding Learning"☆64Dec 8, 2025Updated 2 months ago
- FlashTile is a CUDA Tile IR compiler that is compatible with NVIDIA's tileiras, targeting SM70 through SM121 NVIDIA GPUs.☆54Feb 6, 2026Updated 3 weeks ago