New
kindlingai/gr8 — Context-Parallel KV Added for GLM-5.3-Flash
The latest commit in kindlingai/gr8 adds context-parallel KV to the GLM-5.3-Flash build, storing KV cache pages across multiple ranks, but it is still at an experimental stage.
kindlingai/gr8
The latest commit in kindlingai/gr8 adds context-parallel KV to the GLM-5.3-Flash build, storing KV cache pages across multiple ranks, but it is still at an experimental stage.
View on GitHub ↗- stars
- ★ 35
- Forks
- 9
- Language
- Cuda
- License
- MIT
Posts
1No matching stories.