fuse allreduce and residual_rmsnorm (#8731)

This commit is contained in:
Xiaoyu Zhang
2025-08-11 13:50:53 -07:00
committed by GitHub
parent 8c07fabda7
commit 44e86480e8
8 changed files with 135 additions and 59 deletions
+1 -1
View File
@@ -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",