EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Fix edge case for truncated (#1747)

Browse Source
This commit is contained in:
Byron Hsu
2024-10-22 21:00:25 -07:00
committed by GitHub
parent 1f26e8b8e4
commit 17536e7e3d
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/sglang/srt/managers/scheduler.py
View File

@@ -416,7 +416,7 @@ class Scheduler:
) )
# Truncate prompts that are too long # Truncate prompts that are too long
if len(req.origin_input_ids) >= self.max_req_input_len: if len(req.origin_input_ids) > self.max_req_input_len:
logger.warning( logger.warning(
"Request length is longer than the KV cache pool size or " "Request length is longer than the KV cache pool size or "
"the max context length. Truncated!!!" "the max context length. Truncated!!!"