feat: support glm4 tuning (#8473)

1466c1b8 · Yineng Zhang · GitHub · 9c138a04 · 1466c1b8
Unverified Commit 1466c1b8 authored Jul 28, 2025 by Yineng Zhang Committed by GitHub Jul 28, 2025
Hide whitespace changes
Inline Side-by-side

Showing with 5 additions and 0 deletions

benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py ...hmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py +5 -0

No files found.
--- a/benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py
+++ b/benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py
@@ -427,6 +427,11 @@ def main(args: argparse.Namespace):
        topk = config.num_experts_per_tok
        intermediate_size = config.moe_intermediate_size
        shard_intermediate_size = 2 * intermediate_size // args.tp_size
+    elif config.architectures[0] in ["Glm4MoeForCausalLM"]:
+        E = config.n_routed_experts
+        topk = config.num_experts_per_tok
+        intermediate_size = config.moe_intermediate_size
+        shard_intermediate_size = 2 * intermediate_size // args.tp_size
    else:
        # Default: Mixtral
        E = config.num_local_experts