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 17, 08:00 UTC
Fix log message in scheduler (#652)
Lily Liu committed
Aug 2, 2023 at 13:35 UTC
20044cab7aa6e884e13460506b0e0b6a12722b5d
1 file changed
+5
-5
vllm/core/scheduler.py
+5
-5
@@ -190,13 +190,13 @@ class Scheduler:
190
break
191
192
num_prompt_tokens = seq_group.get_seqs()[0].get_len()
193
- if num_prompt_tokens > min(
194
- self.scheduler_config.max_model_len,
195
- self.scheduler_config.max_num_batched_tokens):
193
+ prompt_limit = min(
194
+ self.scheduler_config.max_model_len,
195
+ self.scheduler_config.max_num_batched_tokens)
196
+ if num_prompt_tokens > prompt_limit:
197
logger.warning(
198
f"Input prompt ({num_prompt_tokens} tokens) is too long"
198
- " and exceeds limit of "
199
- f"{self.scheduler_config.max_model_len}")
199
+ f" and exceeds limit of {prompt_limit}")
200
for seq in seq_group.get_seqs():
201
seq.status = SequenceStatus.FINISHED_IGNORED
202
ignored_seq_groups.append(seq_group)