Fix the default chunked prefill size (#2268)

This commit is contained in:
Lianmin Zheng
2024-11-29 16:03:32 -08:00
committed by GitHub
parent 262e370f78
commit 94e167ea5a
3 changed files with 29 additions and 20 deletions

View File

@@ -253,6 +253,8 @@ class Scheduler:
# Init chunked prefill
self.chunked_prefill_size = server_args.chunked_prefill_size
if self.chunked_prefill_size <= 0: # -1 means disable
self.chunked_prefill_size = None
self.being_chunked_req = None
self.is_mixed_chunk = (
self.chunked_prefill_size is not None and server_args.enable_mixed_chunk