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 10, 04:00 UTC
rename tokenize serving api request id prefix to tokenize (#32328)
Signed-off-by: Andy Xie <andy.xning@gmail.com>
Ning Xie committed
Jan 14, 2026 at 22:52 UTC
552b26293621e586687a523970f554746a5db387
2 files changed
+4
-4
vllm/entrypoints/serve/tokenize/protocol.py
+2
-2
@@ -92,11 +92,11 @@ class TokenizeChatRequest(OpenAIBaseModel):
92
)
93
mm_processor_kwargs: dict[str, Any] | None = Field(
94
default=None,
95
- description=("Additional kwargs to pass to the HF processor."),
95
+ description="Additional kwargs to pass to the HF processor.",
96
)
97
tools: list[ChatCompletionToolsParam] | None = Field(
98
default=None,
99
- description=("A list of tools the model may call."),
99
+ description="A list of tools the model may call.",
100
)
101
102
@model_validator(mode="before")
vllm/entrypoints/serve/tokenize/serving.py
+2
-2
@@ -62,7 +62,7 @@ class OpenAIServingTokenization(OpenAIServing):
62
if error_check_ret is not None:
63
return error_check_ret
64
65
- request_id = f"tokn-{self._base_request_id(raw_request)}"
65
+ request_id = f"tokenize-{self._base_request_id(raw_request)}"
66
67
try:
68
lora_request = self._maybe_get_adapters(request)
@@ -134,7 +134,7 @@ class OpenAIServingTokenization(OpenAIServing):
134
if error_check_ret is not None:
135
return error_check_ret
136
137
- request_id = f"tokn-{self._base_request_id(raw_request)}"
137
+ request_id = f"tokenize-{self._base_request_id(raw_request)}"
138
139
lora_request = self._maybe_get_adapters(request)
140