fuse allreduce and residual_rmsnorm (#8731)

2025-08-12 04:50:53 +08:00
parent 8c07fabda7
commit 44e86480e8
8 changed files with 135 additions and 59 deletions
--- a/python/sglang/srt/server_args.py
+++ b/python/sglang/srt/server_args.py
@@ -1435,7 +1435,7 @@ class ServerArgs:
        parser.add_argument(
            "--enable-flashinfer-allreduce-fusion",
            action="store_true",
-            help="Enable FlashInfer allreduce fusion for Add_RMSNorm.",
+            help="Enable FlashInfer allreduce fusion with Residual RMSNorm.",
        )
        parser.add_argument(
            "--deepep-mode",