Add 320 dimension size support to MLA (#36161)

Signed-off-by: Julien Denize <julien.denize@mistral.ai>
2026-03-11 18:21:22 +01:00
parent 5efa206a8c
commit a5d06dc557
3 changed files with 25 additions and 9 deletions
--- a/vllm/model_executor/layers/attention/mla_attention.py
+++ b/vllm/model_executor/layers/attention/mla_attention.py
@@ -1148,7 +1148,7 @@ class MLACommonBackend(AttentionBackend):

    @classmethod
    def get_supported_head_sizes(cls) -> list[int]:
-        return [576]
+        return [320, 576]

    @classmethod
    def is_mla(cls) -> bool: