ReleaseNVIDIANVIDIApublished Sep 2, 2025seen 2d

NVIDIA/tilus v0.1.1

NVIDIA/tilus

Open original ↗

Captured source

source ↗
published Sep 2, 2025seen 2dcaptured 11hhttp 200method plain

v0.1.1

Repository: NVIDIA/tilus

Tag: v0.1.1

Published: 2025-09-02T19:45:59Z

Prerelease: no

Release notes: This is a small enhancement release of Tilus.

Highlights

  • Add more examples: flash attention with kv-cache, flash linear attention deocde
  • Fix a bug when multiple tilus process access the dispatch table in cache
  • Add targets sm_100, sm_103, sm_110, sm_120 and sm_121.

What's Changed

  • [Docs] Update README.md by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/11
  • [CI] Use RTX 4090 for docs building by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/12
  • [Docs] Update README.md by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/13
  • [Package] Rename to under @NVIDIA organization by @nekomeowww in https://github.com/NVIDIA/tilus/pull/15
  • [Docs] Update installation guide by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/17
  • [CI] Fix concurrency issue by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/18
  • [Docs] Correct gflops to tflops in examples by @YichengDWu in https://github.com/NVIDIA/tilus/pull/19
  • [Example] Add the attention example with kv-cache by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/21
  • [Example] Add example for decoding kernel of flash linear attention by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/25
  • [Example] Add a kernel in the flash linear attention by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/26
  • [Example] Add the fused kernel for decoding of flash linear attention by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/27
  • [Tuning] Add lock to cache dir when dump the tuning result by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/28
  • [Target] Add targets properties by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/29
  • [Bump] Bump version of hidet from 0.6.0 to 0.6.1 by @yaoyaoding in https://github.com/NVIDIA/tilus/pull/30

New Contributors

  • @nekomeowww made their first contribution in https://github.com/NVIDIA/tilus/pull/15
  • @YichengDWu made their first contribution in https://github.com/NVIDIA/tilus/pull/19

Full Changelog: https://github.com/NVIDIA/tilus/compare/v0.1...v0.1.1