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

Fix prefill oom (#1743)

Browse Source
This commit is contained in:
Liangsheng Yin
2024-10-21 03:54:35 -07:00
committed by GitHub
parent 09603c6dc9
commit efb099cdee
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

@@ -427,7 +427,7 @@ class Scheduler:
if req.sampling_params.max_new_tokens is not None if req.sampling_params.max_new_tokens is not None
else 1 << 30 else 1 << 30
), ),
self.max_req_input_len - 1 - len(req.origin_input_ids), self.max_req_input_len - len(req.origin_input_ids),
) )
self.waiting_queue.append(req) self.waiting_queue.append(req)