feat:修复dsa的mqa接口兼容glm5

66979358 · liuchy5 · f7461a96 · 66979358
Commit 66979358 authored Mar 11, 2026 by liuchy5
Show whitespace changes
Inline Side-by-side

Showing with 8 additions and 7 deletions

vllm/model_executor/layers/sparse_attn_indexer.py vllm/model_executor/layers/sparse_attn_indexer.py +8 -7

No files found.
--- a/vllm/model_executor/layers/sparse_attn_indexer.py
+++ b/vllm/model_executor/layers/sparse_attn_indexer.py
@@ -121,9 +121,10 @@ def sparse_attn_indexer(
                    chunk.cu_seqlen_ke,
                    q_fp8[chunk.token_start:chunk.token_end].shape[0],
                    k.shape[0],
-                    64,
+                    q_fp8.shape[1],
-                    128,
+                    q_fp8.shape[2],
-                    True,
+                    None,
+                    True
                    )
            num_rows = logits.shape[0]