From efb099cdee90b9ad332fcda96d89dd91ddebe072 Mon Sep 17 00:00:00 2001 From: Liangsheng Yin Date: Mon, 21 Oct 2024 03:54:35 -0700 Subject: [PATCH] Fix prefill oom (#1743) --- python/sglang/srt/managers/scheduler.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/python/sglang/srt/managers/scheduler.py b/python/sglang/srt/managers/scheduler.py index 990fbeaa8..1b68bacd9 100644 --- a/python/sglang/srt/managers/scheduler.py +++ b/python/sglang/srt/managers/scheduler.py @@ -427,7 +427,7 @@ class Scheduler: if req.sampling_params.max_new_tokens is not None 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)