iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Enable DeepSeek V4 and GLM-5.1 on SM120 (#43477)

Browse Source 
Signed-off-by: Zihua Wu <13583761+lucifer1004@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
This commit is contained in:
Gabriel Wu
2026-06-22 11:54:14 -07:00
committed by GitHub
co-authored by Claude Opus 4.8 Yongye Zhu
parent 3ce15fd574
commit 44d95069e9
37 changed files with 2347 additions and 476 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/pre_commit/generate_attention_backend_docs.py
+5 -2
View File
@@ -690,7 +690,9 @@ def parse_compute_capability(node: ast.ClassDef) -> str:
major_list.sort()
if len(major_list) == 1:
return f"{major_list[0]}.x"
return f"{major_list[0]}.x-{major_list[-1]}.x"
if major_list == list(range(major_list[0], major_list[-1] + 1)):
return f"{major_list[0]}.x-{major_list[-1]}.x"
return ", ".join(f"{major}.x" for major in major_list)
if min_cap:
if max_cap:
@@ -1668,7 +1670,8 @@ def generate_mla_section(
"`--attention-backend=<BACKEND>` (e.g., `FLASHMLA_SPARSE_DSV4`,",
"`FLASHINFER_MLA_SPARSE_DSV4`). They share the V4 sparse-index",
"pipeline (compressor + SWA + indexer, 256-token blocks, head 512);",
"default on NVIDIA is `FLASHMLA_SPARSE_DSV4`.",
"default on NVIDIA is `FLASHINFER_MLA_SPARSE_DSV4` on SM12x and",
"`FLASHMLA_SPARSE_DSV4` on other supported CUDA architectures.",
"",
]
)