- 01 Aug, 2025 8 commits
- 31 Jul, 2025 8 commits
- 30 Jul, 2025 4 commits
- 29 Jul, 2025 4 commits
- 28 Jul, 2025 3 commits
- 26 Jul, 2025 1 commit
-
-
yangql authored
-
- 25 Jul, 2025 6 commits
- 24 Jul, 2025 4 commits
- 22 Jul, 2025 2 commits
[fix]避免mla中cudagraph的适配影响非并行解码的逻辑 See merge request dcutoolkit/deeplearing/vllm!165
[feat]支持v1 engine mtp cudagraph See merge request dcutoolkit/deeplearing/vllm!164
Merge v0.9.2-dev-disagg into v0.9.2-dev See merge request dcutoolkit/deeplearing/vllm!163
[fix]解决deepseek报错 See merge request dcutoolkit/deeplearing/vllm!162
[Fix] MLA only supports decode-only full CUDAGraph capture. Make sure all cudagraph capture sizes <= max_num_seq.