🚀🚀 Efficient implementations of Native Sparse Attention
☆622Sep 29, 2025Updated 10 months ago
Alternatives and similar repositories for Flash-Sparse-Attention
Users that are interested in Flash-Sparse-Attention are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- ☆32Jul 2, 2025Updated last year
- ☆253Nov 19, 2025Updated 8 months ago
- An experimental communicating attention kernel based on DeepEP.☆34Jul 29, 2025Updated last year
- An efficient implementation of the NSA (Native Sparse Attention) kernel☆134Jun 24, 2025Updated last year
- ☆48Dec 13, 2025Updated 7 months ago
- Managed Database hosting by DigitalOcean • AdPostgreSQL, MySQL, MongoDB, Kafka, Valkey, and OpenSearch available. Automatically scale up storage and focus on building your apps.
- Low overhead tracing library and trace visualizer for pipelined CUDA kernels☆135Jul 29, 2026Updated last week
- FA4-based Relative Attention Kernel developed by TML and Colfax☆18Jul 17, 2026Updated 3 weeks ago
- 🐳 Efficient Triton implementations for "Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention"☆1,018Feb 5, 2026Updated 6 months ago
- CUDA kernels for linear attention variants, written in CuTe DSL and CUTLASS C++.☆537Updated this week
- ☆37Aug 7, 2025Updated last year
- [ICML2025] SpargeAttention: A training-free sparse attention that accelerates any model inference.☆1,025Feb 25, 2026Updated 5 months ago
- A collection of specialized agent skills for AI infrastructure development, enabling Claude Code to write, optimize, and debug high-perfo…☆142Jul 9, 2026Updated last month
- Tilus is a tile-level kernel programming language with explicit control over shared memory and registers.☆490Jul 5, 2026Updated last month
- FLA but cuTile☆27Apr 17, 2026Updated 3 months ago
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- NVSHMEM‑Tutorial: Build a DeepEP‑like GPU Buffer☆196Feb 11, 2026Updated 6 months ago
- mKernel: fast multi-node, multi-GPU fused kernels☆264Updated this week
- A Survey of Efficient Attention Methods: Hardware-efficient, Sparse, Compact, and Linear Attention☆307Dec 1, 2025Updated 8 months ago
- ☆47Jul 27, 2026Updated 2 weeks ago
- ☆125May 19, 2025Updated last year
- High-performance LLM operator library built on TileLang.☆168Updated this week
- DeeperGEMM: crazy optimized version☆86May 5, 2025Updated last year
- A Quirky Assortment of CuTe Kernels☆1,101Updated this week
- A Distributed Attention Towards Linear Scalability for Ultra-Long Context, Heterogeneous Data Training☆907Updated this week
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- NVIDIA cuTile learn☆169Dec 9, 2025Updated 8 months ago
- Distributed Compiler based on Triton for Parallel Systems☆1,515Updated this week
- Compact and Agent-Native MoE Training System☆328Jul 31, 2026Updated last week
- high-performance linear attention kernel library built on TileLang☆629Updated this week
- Vortex: Programmable Sparse Attention for Agents as Algorithm Designers☆68Updated this week
- Tile-based language built for AI computation across all scales☆185Updated this week
- SeerAttention: Learning Intrinsic Sparse Attention in Your LLMs☆216Jul 10, 2026Updated last month
- Benchmark tests supporting the TiledCUDA library.☆19Nov 19, 2024Updated last year
- Official repository Flash Local Linear Attention☆38May 28, 2026Updated 2 months ago
- Managed hosting for WordPress and PHP on Cloudways • AdManaged hosting for WordPress, Magento, Laravel, or PHP apps, on multiple cloud providers. Deploy in minutes on Cloudways by DigitalOcean.
- Perplexity GPU Kernels☆601Nov 7, 2025Updated 9 months ago
- [MLSys 26] 🥇 Solution for Gated Delta Net Track of MLSys 26 Flash infer competition☆36May 22, 2026Updated 2 months ago
- FlashTile is a CUDA Tile IR compiler that is compatible with NVIDIA's tileiras, targeting SM70 through SM121 NVIDIA GPUs.☆61Feb 6, 2026Updated 6 months ago
- FlashKDA: high-performance Kimi Delta Attention kernels☆1,205Jul 30, 2026Updated last week
- Accelerating MoE with IO and Tile-aware Optimizations☆739Jul 4, 2026Updated last month
- ☆35Dec 31, 2025Updated 7 months ago
- A sparse attention kernel supporting mix sparse patterns☆545Jan 18, 2026Updated 6 months ago