Devotd Github

pentacoxian/qwen-qsa-llama.cpp

pentacoxian/qwen-qsa-llama.cpp 是一个 llama.cpp 分支,通过应用 Qwen Sparse Attention,让 Qwen3.8-27B 支持长上下文与图像输入。

GitHub 项目说明: Qwen3.8-27B with Qwen Sparse Attention on llama.cpp: 262K context with vision on one 24 GB GPU

在 GitHub 查看 ↗
星标
★ 3
复刻
0
语言
C++
许可证
MIT

文章

1