From e1ce44cdb1e427c8dcbfd42f7181d08be8484b15 Mon Sep 17 00:00:00 2001 From: Swipe4057 <106391009+Swipe4057@users.noreply.github.com> Date: Sat, 7 Jun 2025 13:06:58 +0300 Subject: [PATCH] Disabling mixed chunked prefill when eagle is enabled (#6874) --- python/sglang/srt/server_args.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/python/sglang/srt/server_args.py b/python/sglang/srt/server_args.py index 317df60a6..99928f1b7 100644 --- a/python/sglang/srt/server_args.py +++ b/python/sglang/srt/server_args.py @@ -425,6 +425,12 @@ class ServerArgs: "Overlap scheduler is disabled because of using " "eagle speculative decoding." ) + if self.enable_mixed_chunk: + self.enable_mixed_chunk = False + logger.warning( + "Mixed chunked prefill is disabled because of using " + "eagle speculative decoding." + ) model_arch = get_model_arch(self)