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 8, 08:45 UTC
Fix num_logprobs parameter description in sampler.py (#34451)
Signed-off-by: Zhuohan Li <zhuohan123@gmail.com>
Zhuohan Li committed
Feb 12, 2026 at 18:18 UTC
d707678dfb9a1f616d174022ebc74065d1011863
1 file changed
+1
-1
vllm/v1/sample/sampler.py
+1
-1
@@ -217,7 +217,7 @@ class Sampler(nn.Module):
217
218
Args:
219
logprobs: (num tokens) x (vocab) tensor
220
- num_logprobs: minimum number of logprobs to
220
+ num_logprobs: maximum number of logprobs to
221
retain per token
222
token_ids: prompt tokens (if prompt logprobs)
223
or sampled tokens (if sampled