diff --git a/python/sglang/srt/managers/scheduler.py b/python/sglang/srt/managers/scheduler.py index 05bc8d730..cb3a4b5de 100644 --- a/python/sglang/srt/managers/scheduler.py +++ b/python/sglang/srt/managers/scheduler.py @@ -957,7 +957,11 @@ class Scheduler: self.req_to_token_pool.free(self.chunked_req.req_pool_idx) self.batch_is_full = False + last_bs = self.last_batch.batch_size() self.last_batch.filter_batch() + if self.last_batch.batch_size() < last_bs: + self.batch_is_full = False + if not self.last_batch.is_empty(): if self.running_batch is None: self.running_batch = self.last_batch