[Fix] DeepEP Compatibility with Low Latency (#5068)

Co-authored-by: ch-wan <cwan39@gatech.edu>
2025-04-09 11:31:31 +08:00
parent aac531c53b
commit bc3f6db2dd
4 changed files with 146 additions and 118 deletions
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -1101,6 +1101,7 @@ class ServerArgs:
            "--deepep-mode",
            type=str,
            choices=["normal", "low_latency", "auto"],
+            default="auto",
            help="Select the mode when enable DeepEP MoE, could be `normal`, `low_latency` or `auto`. Default is `auto`, which means `low_latency` for decode batch and `normal` for prefill batch.",
        )