[tiny] Remove unsupported TRITON_MLA backend from batch invariance (#28832)

Signed-off-by: Bram Wasti <[email protected]>
Signed-off-by: Bram Wasti <[email protected]>
Co-authored-by: Wentao Ye <[email protected]>

Author: bwasti

vllm/model_executor/layers/batch_invariant.py

@@ -805,11 +805,11 @@ def override_envs_for_invariance():
         "FLASH_ATTN",  # best supported backend
         "FLASHINFER",
         "FLASH_ATTN_MLA",
-        "TRITON_MLA",
         # Not yet supported MLA backends
         # "FLASHMLA",
         # "FLEX_ATTENTION", # IMA issue even if we disable batch invariance
         # "FLASHINFER_MLA", https://github.com/vllm-project/vllm/pull/28967
+        # "TRITON_MLA",
     ]
     if curr_attn_backend not in supported_backends:
         warning = (