xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 01:45:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Model] Prioritize Transformers fallback over suffix matching (#21719)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung 2025-07-28 17:15:31 +08:00 committed by GitHub
parent d128d0d554
commit a4ed731546
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 38 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
tests/models/multimodal/generation/test_common.py
View File

@ -222,6 +222,7 @@ VLM_TEST_SETTINGS = {
},
marks=[large_gpu_mark(min_gb=32)],
),
# Check "auto" with fallback to transformers
"internvl-transformers": VLMTestInfo(
models=["OpenGVLab/InternVL3-1B-hf"],
test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),
@ -231,7 +232,7 @@ VLM_TEST_SETTINGS = {
use_tokenizer_eos=True,
image_size_factors=[(0.25, 0.5, 1.0)],
vllm_runner_kwargs={
"model_impl": "transformers",
"model_impl": "auto",
},
auto_cls=AutoModelForImageTextToText,
marks=[pytest.mark.core_model],

58
vllm/model_executor/models/registry.py
View File

@ -586,18 +586,6 @@ class _ModelRegistry:
return architecture
def _normalize_archs(
self,
architectures: list[str],
model_config: ModelConfig,
) -> list[str]:
if not architectures:
logger.warning("No model architectures are specified")
return [
self._normalize_arch(arch, model_config) for arch in architectures
]
def inspect_model_cls(
self,
architectures: Union[str, list[str]],
@ -605,8 +593,8 @@ class _ModelRegistry:
) -> tuple[_ModelInfo, str]:
if isinstance(architectures, str):
architectures = [architectures]
normalized_archs = self._normalize_archs(architectures, model_config)
if not architectures:
raise ValueError("No model architectures are specified")
# Require transformers impl
if model_config.model_impl == ModelImpl.TRANSFORMERS:
@ -617,13 +605,26 @@ class _ModelRegistry:
if model_info is not None:
return (model_info, arch)
for arch, normalized_arch in zip(architectures, normalized_archs):
# Fallback to transformers impl (after resolving convert_type)
if (all(arch not in self.models for arch in architectures)
and model_config.model_impl == ModelImpl.AUTO
and getattr(model_config, "convert_type", "none") == "none"):
arch = self._try_resolve_transformers(architectures[0],
model_config)
if arch is not None:
model_info = self._try_inspect_model_cls(arch)
if model_info is not None:
return (model_info, arch)
for arch in architectures:
normalized_arch = self._normalize_arch(arch, model_config)
model_info = self._try_inspect_model_cls(normalized_arch)
if model_info is not None:
return (model_info, arch)
# Fallback to transformers impl
if model_config.model_impl in (ModelImpl.AUTO, ModelImpl.TRANSFORMERS):
# Fallback to transformers impl (before resolving runner_type)
if (all(arch not in self.models for arch in architectures)
and model_config.model_impl == ModelImpl.AUTO):
arch = self._try_resolve_transformers(architectures[0],
model_config)
if arch is not None:
@ -640,8 +641,8 @@ class _ModelRegistry:
) -> tuple[type[nn.Module], str]:
if isinstance(architectures, str):
architectures = [architectures]
normalized_archs = self._normalize_archs(architectures, model_config)
if not architectures:
raise ValueError("No model architectures are specified")
# Require transformers impl
if model_config.model_impl == ModelImpl.TRANSFORMERS:
@ -652,13 +653,26 @@ class _ModelRegistry:
if model_cls is not None:
return (model_cls, arch)
for arch, normalized_arch in zip(architectures, normalized_archs):
# Fallback to transformers impl (after resolving convert_type)
if (all(arch not in self.models for arch in architectures)
and model_config.model_impl == ModelImpl.AUTO
and getattr(model_config, "convert_type", "none") == "none"):
arch = self._try_resolve_transformers(architectures[0],
model_config)
if arch is not None:
model_cls = self._try_load_model_cls(arch)
if model_cls is not None:
return (model_cls, arch)
for arch in architectures:
normalized_arch = self._normalize_arch(arch, model_config)
model_cls = self._try_load_model_cls(normalized_arch)
if model_cls is not None:
return (model_cls, arch)
# Fallback to transformers impl
if model_config.model_impl in (ModelImpl.AUTO, ModelImpl.TRANSFORMERS):
# Fallback to transformers impl (before resolving runner_type)
if (all(arch not in self.models for arch in architectures)
and model_config.model_impl == ModelImpl.AUTO):
arch = self._try_resolve_transformers(architectures[0],
model_config)
if arch is not None: