A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 17:15 UTC
Fix Transformers backend compatibility check (#15290)
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Harry Mellor committed
Mar 31, 2025 at 18:27 UTC
d4bfc23ef0319e2af6b01a715572364d947aee29
1 file changed
+1
-4
vllm/model_executor/model_loader/utils.py
+1
-4
@@ -36,10 +36,7 @@ def is_transformers_impl_compatible(
36
mod = module or getattr(transformers, arch, None)
37
if mod is None:
38
return False
39
- if hasattr(mod, "supports_backend"):
40
- return mod.is_backend_compatible()
41
- else:
42
- return mod._supports_flex_attn
39
+ return mod.is_backend_compatible()
40
41
42
def resolve_transformers_fallback(model_config: ModelConfig,