[Fix] Move "model_config" as keyword args in chat_utils.py (#18098)

Signed-off-by: Linkun <github@lkchen.net>

vllm/entrypoints/chat_utils.py

@@ -44,7 +44,7 @@ from vllm.transformers_utils.chat_templates import (
 # yapf: enable
 from vllm.transformers_utils.processor import cached_get_processor
 from vllm.transformers_utils.tokenizer import AnyTokenizer, MistralTokenizer
-from vllm.utils import random_uuid
+from vllm.utils import deprecate_kwargs, random_uuid
 
 logger = init_logger(__name__)
 
@@ -329,11 +329,17 @@ def resolve_mistral_chat_template(
             "so it will be ignored.")
     return None
 
+@deprecate_kwargs(
+    "trust_remote_code",
+    additional_message="Please use `model_config.trust_remote_code` instead.",
+)
 def resolve_hf_chat_template(
-    model_config: ModelConfig,
     tokenizer: Union[PreTrainedTokenizer, PreTrainedTokenizerFast],
     chat_template: Optional[str],
     tools: Optional[list[dict[str, Any]]],
+    *,
+    model_config: ModelConfig,
+    trsut_remote_code: Optional[bool] = None,
 ) -> Optional[str]:
     # 1st priority: The given chat template
     if chat_template is not None:
@@ -379,18 +385,19 @@ def resolve_hf_chat_template(
 
 
 def _resolve_chat_template_content_format(
-    model_config: ModelConfig,
     chat_template: Optional[str],
     tools: Optional[list[dict[str, Any]]],
     given_format: ChatTemplateContentFormatOption,
     tokenizer: AnyTokenizer,
+    *,
+    model_config: ModelConfig,
 ) -> _ChatTemplateContentFormat:
     if isinstance(tokenizer, (PreTrainedTokenizer, PreTrainedTokenizerFast)):
         hf_chat_template = resolve_hf_chat_template(
-            model_config,
             tokenizer,
             chat_template=chat_template,
             tools=tools,
+            model_config=model_config,
         )
     else:
         hf_chat_template = None
@@ -428,19 +435,25 @@ def _log_chat_template_content_format(
         )
 
 
+@deprecate_kwargs(
+    "trust_remote_code",
+    additional_message="Please use `model_config.trust_remote_code` instead.",
+)
 def resolve_chat_template_content_format(
-    model_config: ModelConfig,
     chat_template: Optional[str],
     tools: Optional[list[dict[str, Any]]],
     given_format: ChatTemplateContentFormatOption,
     tokenizer: AnyTokenizer,
+    *,
+    model_config: ModelConfig,
+    trust_remote_code: Optional[bool] = None,
 ) -> _ChatTemplateContentFormat:
     detected_format = _resolve_chat_template_content_format(
-        model_config,
         chat_template,
         tools,
         given_format,
         tokenizer,
+        model_config=model_config,
     )
 
     _log_chat_template_content_format(
@@ -1191,21 +1204,27 @@ def parse_chat_messages_futures(
     return conversation, mm_tracker.all_mm_data()
 
 
+@deprecate_kwargs(
+    "trust_remote_code",
+    additional_message="Please use `model_config.trust_remote_code` instead.",
+)
 def apply_hf_chat_template(
-    model_config: ModelConfig,
     tokenizer: Union[PreTrainedTokenizer, PreTrainedTokenizerFast],
     conversation: list[ConversationMessage],
     chat_template: Optional[str],
     tools: Optional[list[dict[str, Any]]],
     *,
+    model_config: ModelConfig,
     tokenize: bool = False,  # Different from HF's default
+    # Deprecated, explicitly capture here so it doesn't slit into kwargs.
+    trust_remote_code: Optional[bool] = None,
     **kwargs: Any,
 ) -> str:
     hf_chat_template = resolve_hf_chat_template(
-        model_config,
         tokenizer,
         chat_template=chat_template,
         tools=tools,
+        model_config=model_config,
     )
 
     if hf_chat_template is None: