biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Bugfix]: do not shutdown server if skip_special_use=False for MistralTokenizer (#14094)

Browse Source 
Signed-off-by: Guillaume Calmettes <gcalmettes@scaleway.com>
This commit is contained in:
Guillaume Calmettes
2025-04-10 04:43:09 +02:00
committed by GitHub
parent 1e44ffc3ff
commit 1da6a09274
3 changed files with 13 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/entrypoints/openai/serving_chat.py
View File

@@ -39,7 +39,8 @@ from vllm.sampling_params import BeamSearchParams, SamplingParams
from vllm.sequence import Logprob from vllm.sequence import Logprob
from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer
from vllm.transformers_utils.tokenizers import (maybe_serialize_tool_calls, from vllm.transformers_utils.tokenizers import (maybe_serialize_tool_calls,
truncate_tool_call_ids) truncate_tool_call_ids,
validate_request_params)
logger = init_logger(__name__) logger = init_logger(__name__)
@@ -159,6 +160,7 @@ class OpenAIServingChat(OpenAIServing):
# for more info: see comment in `maybe_serialize_tool_calls` # for more info: see comment in `maybe_serialize_tool_calls`
maybe_serialize_tool_calls(request) maybe_serialize_tool_calls(request)
truncate_tool_call_ids(request) truncate_tool_call_ids(request)
validate_request_params(request)
if (request.tool_choice == "auto" and if (request.tool_choice == "auto" and
not (self.enable_auto_tools and tool_parser is not None) not (self.enable_auto_tools and tool_parser is not None)

5
vllm/transformers_utils/tokenizers/__init__.py
View File

@@ -1,8 +1,9 @@
# SPDX-License-Identifier: Apache-2.0 # SPDX-License-Identifier: Apache-2.0
from .mistral import (MistralTokenizer, maybe_serialize_tool_calls, from .mistral import (MistralTokenizer, maybe_serialize_tool_calls,
truncate_tool_call_ids) truncate_tool_call_ids, validate_request_params)
__all__ = [ __all__ = [
"MistralTokenizer", "maybe_serialize_tool_calls", "truncate_tool_call_ids" "MistralTokenizer", "maybe_serialize_tool_calls", "truncate_tool_call_ids",
"validate_request_params"
] ]

7
vllm/transformers_utils/tokenizers/mistral.py
View File

@@ -98,6 +98,13 @@ def truncate_tool_call_ids(request: "ChatCompletionRequest"):
request.messages[i]["tool_call_id"] = tool_call_id request.messages[i]["tool_call_id"] = tool_call_id
def validate_request_params(request: "ChatCompletionRequest"):
if (request.skip_special_tokens is not None
and not request.skip_special_tokens):
raise ValueError("skip_special_tokens=False is not supported "
"for Mistral tokenizers.")
def list_local_repo_files(repo_id: str, revision: Optional[str]) -> List[str]: def list_local_repo_files(repo_id: str, revision: Optional[str]) -> List[str]:
repo_cache = os.path.join( repo_cache = os.path.join(
huggingface_hub.constants.HF_HUB_CACHE, huggingface_hub.constants.HF_HUB_CACHE,