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 16, 19:30 UTC
[Misc] improve logits processors logging message (#7435)
Andrew Wang committed
Aug 12, 2024 at 19:29 UTC
97a6be95ba279c2bc9cdd8890506ec94e63b268d
2 files changed
+1
-2
tests/entrypoints/openai/test_serving_chat.py
-1
@@ -73,7 +73,6 @@ def test_serving_chat_should_set_correct_max_tokens():
73
with suppress(Exception):
74
asyncio.run(serving_chat.create_chat_completion(req))
75
76
- # AsyncLLMEngine.generate(inputs, sampling_params, ...)
76
assert mock_engine.generate.call_args.args[1].max_tokens == 93
77
78
req.max_tokens = 10
vllm/entrypoints/openai/logits_processors.py
+1
-1
@@ -71,7 +71,7 @@ def get_logits_processors(
71
# Check if token_id is within the vocab size
72
for token_id, bias in clamped_logit_bias.items():
73
if token_id < 0 or token_id >= tokenizer.vocab_size:
74
- raise ValueError("token_id in logit_bias contains "
74
+ raise ValueError(f"token_id {token_id} in logit_bias contains "
75
"out-of-vocab token id")
76
77
logits_processors.append(