DeepSeek Native Sparse Attention pytorch implementation
β118Dec 17, 2025Updated 5 months ago
Alternatives and similar repositories for NSA-pytorch
Users that are interested in NSA-pytorch are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- π³ Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"β1,004Feb 5, 2026Updated 4 months ago
- Implementation of the sparse attention pattern proposed by the Deepseek team in their "Native Sparse Attention" paperβ807Aug 15, 2025Updated 9 months ago
- qwen-nsaβ87Oct 14, 2025Updated 7 months ago
- An efficient implementation of the NSA (Native Sparse Attention) kernelβ134Jun 24, 2025Updated 11 months ago
- Complete simulation of IEEE 754 fixed and floating point specification to any precisionβ13Aug 26, 2020Updated 5 years ago
- Bare Metal GPUs on DigitalOcean Gradient AI β’ AdPurpose-built for serious AI teams training foundational models, running large-scale inference, and pushing the boundaries of what's possible.
- Compact and Agent-Native MoE Training Systemβ144Jun 5, 2026Updated last week
- LongAttn οΌSelecting Long-context Training Data via Token-level Attentionβ15Jul 16, 2025Updated 10 months ago
- High performance RMSNorm Implement by using SM Core Storage(Registers and Shared Memory)β30Jan 22, 2026Updated 4 months ago
- Code for paper: [ICLR2025 Oral] FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inferenceβ168Oct 13, 2025Updated 7 months ago
- [ICLR 2025] DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Headsβ541Feb 10, 2025Updated last year
- C++ library for finding Strongly Connected Components in parallel, based on paper: https://dl.acm.org/citation.cfm?id=2851161β12May 22, 2018Updated 8 years ago
- Nex Venus Communication Libraryβ75Nov 17, 2025Updated 6 months ago
- Website for CSE 234, Winter 2025β15Mar 24, 2025Updated last year
- Terminal UI for NVIDIA Nsight Systems profiles β timeline viewer, kernel navigator, NVTX hierarchyβ58Updated this week
- Managed Kubernetes at scale on DigitalOcean β’ AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- DeepSeek-V3.2-Exp DSA Warmup Lightning Indexer training operator based on tilelangβ44Nov 19, 2025Updated 6 months ago
- Expert Specialization MoE Solution based on CUTLASSβ27Apr 14, 2026Updated last month
- TileFusion is an experimental C++ macro kernel template library that elevates the abstraction level in CUDA C for tile processing.β110Jun 28, 2025Updated 11 months ago
- Benchmark tests supporting the TiledCUDA library.β19Nov 19, 2024Updated last year
- [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.β1,002Feb 25, 2026Updated 3 months ago
- TileGraph is an experimental DNN compiler that utilizes static code generation and kernel fusion techniques.β11Sep 18, 2024Updated last year
- β139May 29, 2025Updated last year
- Algorithms for approximate attention in LLMsβ22Apr 14, 2025Updated last year
- Codebase for decoding compressed trust.β27May 7, 2024Updated 2 years ago
- Managed Kubernetes at scale on DigitalOcean β’ AdDigitalOcean Kubernetes includes the control plane, bandwidth allowance, container registry, automatic updates, and more for free.
- β71Feb 5, 2026Updated 4 months ago
- CUDA SGEMM optimization noteβ15Oct 31, 2023Updated 2 years ago
- Efficient LLM Inference over Long Sequencesβ394Jun 25, 2025Updated 11 months ago
- A collection of memory efficient attention operators implemented in the Triton language.β295Jun 1, 2026Updated last week
- Official Implementation of our ICML 2025 paper: "D-MoLE: Dynamic Mixture of Curriculum LoRA Experts for Continual Multimodal Instruction β¦β27Jan 11, 2026Updated 5 months ago
- FlyDSL is the Python frontβend of the project: Flexible LaYout DSL.β198Updated this week
- analyse problems of AI with Math and Codeβ30Jul 28, 2025Updated 10 months ago
- A sparse attention kernel supporting mix sparse patternsβ518Jan 18, 2026Updated 4 months ago
- A simple benchmark for modern image formats on mobiles, including WebP/HEIC/BPG/FLIF/AVIFβ12Oct 31, 2019Updated 6 years ago
- Deploy to Railway using AI coding agents - Free Credits Offer β’ AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- MoBA: Mixture of Block Attention for Long-Context LLMsβ2,123Apr 3, 2025Updated last year
- β282Jun 6, 2025Updated last year
- Vero: An Open RL Recipe for General Visual Reasoningβ122Jun 3, 2026Updated last week
- A lightweight reinforcement learning framework that integrates seamlessly into your codebase, empowering developers to focus on algorithmβ¦β104Aug 25, 2025Updated 9 months ago
- β95Updated this week
- Wave: Python Domain-Specific Language for High Performance Machine Learningβ57Updated this week
- Tile-Based Runtime for Ultra-Low-Latency LLM Inferenceβ1,289Updated this week