[Transformers backend] Find attention with a fuser and attach vLLM's layer to it (#54941)

Signed-off-by: Thomas Ortner <boh@zurich.ibm.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Thomas Ortner
2026-09-04 16:26:01 +00:00
committed by GitHub
co-authored by Harry Mellor
parent 99a1ab8e4c
commit 8ad2076c4a
14 changed files with 521 additions and 150 deletions
+4 -1
View File
@@ -1253,7 +1253,10 @@ class _ModelRegistry:
"'auto_map' (relevant if the model is custom)."
)
if not model_module.is_backend_compatible():
if not (
model_module.is_backend_compatible()
or model_module._can_set_attn_implementation()
):
if model_config.model_impl != "transformers":
return None