xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-26 07:11:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Fix] check to make sure processor has chat templates (#18047)

Browse Source 
Signed-off-by: Aaron Pham <contact@aarnphm.xyz>
This commit is contained in:
Aaron Pham 2025-05-13 06:04:10 -04:00 committed by GitHub
parent 98fcba1575
commit cb528d0585
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tests/compile/test_pass_manager.py
View File

@ -22,7 +22,7 @@ def test_bad_callable():
pass_manager.configure(config) pass_manager.configure(config)
with pytest.raises(AssertionError): with pytest.raises(AssertionError):
pass_manager.add(simple_callable) # noqa, type wrong on purpose pass_manager.add(simple_callable)
# Pass that inherits from InductorPass # Pass that inherits from InductorPass

2
vllm/compilation/inductor_pass.py
View File

@ -16,7 +16,7 @@ if is_torch_equal_or_newer("2.6"):
from torch._inductor.custom_graph_pass import CustomGraphPass from torch._inductor.custom_graph_pass import CustomGraphPass
else: else:
# CustomGraphPass is not present in 2.5 or lower, import our version # CustomGraphPass is not present in 2.5 or lower, import our version
from .torch25_custom_graph_pass import ( # noqa: yapf from .torch25_custom_graph_pass import ( # noqa: E501
Torch25CustomGraphPass as CustomGraphPass) Torch25CustomGraphPass as CustomGraphPass)
_pass_context = None _pass_context = None

4
vllm/entrypoints/chat_utils.py
View File

@ -349,11 +349,11 @@ def resolve_hf_chat_template(
trust_remote_code=model_config.trust_remote_code, trust_remote_code=model_config.trust_remote_code,
) )
if isinstance(processor, ProcessorMixin) and \ if isinstance(processor, ProcessorMixin) and \
hasattr(processor, 'chat_template') and \
processor.chat_template is not None: processor.chat_template is not None:
return processor.chat_template return processor.chat_template
except Exception: except Exception:
logger.debug("Failed to load AutoProcessor chat template for %s", logger.debug("Failed to load AutoProcessor chat template for %s", tokenizer.name_or_path, exc_info=True) # noqa: E501
tokenizer.name_or_path, exc_info=True)
# 3rd priority: AutoTokenizer chat template # 3rd priority: AutoTokenizer chat template
try: try: