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 15, 08:30 UTC
refactor: simplify request handler, use positive condition check for handler assignment (#18690)
Signed-off-by: googs1025 <googs1025@gmail.com>
CYJiang committed
May 26, 2025 at 14:32 UTC
abd4030d94206db4b0ea6c42640e2782ccd62532
1 file changed
+6
-6
vllm/entrypoints/openai/run_batch.py
+6
-6
@@ -365,8 +365,8 @@ async def main(args):
365
366
# Determine the type of request and run it.
367
if request.url == "/v1/chat/completions":
368
- chat_handler_fn = (None if openai_serving_chat is None else
369
- openai_serving_chat.create_chat_completion)
368
+ chat_handler_fn = openai_serving_chat.create_chat_completion if \
369
+ openai_serving_chat is not None else None
370
if chat_handler_fn is None:
371
response_futures.append(
372
make_async_error_request_output(
@@ -380,8 +380,8 @@ async def main(args):
380
run_request(chat_handler_fn, request, tracker))
381
tracker.submitted()
382
elif request.url == "/v1/embeddings":
383
- embed_handler_fn = (None if openai_serving_embedding is None else
384
- openai_serving_embedding.create_embedding)
383
+ embed_handler_fn = openai_serving_embedding.create_embedding if \
384
+ openai_serving_embedding is not None else None
385
if embed_handler_fn is None:
386
response_futures.append(
387
make_async_error_request_output(
@@ -394,8 +394,8 @@ async def main(args):
394
run_request(embed_handler_fn, request, tracker))
395
tracker.submitted()
396
elif request.url == "/v1/score":
397
- score_handler_fn = (None if openai_serving_scores is None else
398
- openai_serving_scores.create_score)
397
+ score_handler_fn = openai_serving_scores.create_score if \
398
+ openai_serving_scores is not None else None
399
if score_handler_fn is None:
400
response_futures.append(
401
make_async_error_request_output(