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, 06:00 UTC
[BugFix] Handle missing sep_token for Qwen3-Reranker in Score API (#19522)
Signed-off-by: strutive07 <strutive07@gmail.com>
wonjun Jang committed
Jun 12, 2025 at 16:00 UTC
1b0b065eb59a8a7751a0ac8fa9ebd182ed627c35
1 file changed
+2
-2
vllm/entrypoints/openai/serving_score.py
+2
-2
@@ -174,8 +174,8 @@ class ServingScores(OpenAIServing):
174
for t1, t2 in input_pairs))
175
176
for prompt_inputs, (t1, t2) in zip(tokenized_prompts, input_pairs):
177
-
178
- request_prompt = f"{t1}{tokenizer.sep_token}{t2}"
177
+ sep_token = tokenizer.sep_token if tokenizer.sep_token else ''
178
+ request_prompt = f"{t1}{sep_token}{t2}"
179
180
input_ids = prompt_inputs["input_ids"]
181
text_token_prompt = \