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 14, 05:30 UTC
[Docs] Fix warnings in mkdocs build (continued) (#25163)
Signed-off-by: Zerohertz <ohg3417@gmail.com>
Hyogeun Oh (오효근) committed
Sep 19, 2025 at 00:23 UTC
b419937c78017dc4c5bfa19f11547f4832ea2290
2 files changed
+3
-7
vllm/distributed/device_communicators/shm_object_storage.py
+1
-1
@@ -253,7 +253,7 @@ class SingleWriterShmRingBuffer:
253
254
Args:
255
nbytes (int, optional): The size of the buffer to free. If None,
256
- frees the maximum size of the ring buffer.
256
+ frees the maximum size of the ring buffer.
257
'''
258
259
assert self.is_writer, "Only the writer can free buffers."
vllm/entrypoints/openai/serving_engine.py
+2
-6
@@ -697,9 +697,7 @@ class OpenAIServing:
697
add_special_tokens: bool = True,
698
) -> TextTokensPrompt:
699
"""
700
- A simpler implementation of
701
- [`_tokenize_prompt_input_or_inputs`][vllm.entrypoints.openai.serving_engine.OpenAIServing._tokenize_prompt_input_or_inputs]
702
- that assumes single input.
700
+ A simpler implementation that tokenizes a single prompt input.
701
"""
702
async for result in self._tokenize_prompt_inputs_async(
703
request,
@@ -718,9 +716,7 @@ class OpenAIServing:
716
add_special_tokens: bool = True,
717
) -> AsyncGenerator[TextTokensPrompt, None]:
718
"""
721
- A simpler implementation of
722
- [`_tokenize_prompt_input_or_inputs`][vllm.entrypoints.openai.serving_engine.OpenAIServing._tokenize_prompt_input_or_inputs]
723
- that assumes multiple inputs.
719
+ A simpler implementation that tokenizes multiple prompt inputs.
720
"""
721
for prompt in prompt_inputs:
722
if isinstance(prompt, str):