A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Bench] Add DeepSeekV32 to MoE benchmark (#25962)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Jee Jee Li committed
Oct 1, 2025 at 05:13 UTC
67f3fb08440e15d6a9c12a72266d8ca60eed5699
1 file changed
+2
-1
benchmarks/kernels/benchmark_moe.py
+2
-1
@@ -584,8 +584,9 @@ def main(args: argparse.Namespace):
584
topk = config.num_experts_per_tok
585
intermediate_size = config.intermediate_size
586
elif config.architectures[0] in (
587
- "DeepseekV3ForCausalLM",
587
"DeepseekV2ForCausalLM",
588
+ "DeepseekV3ForCausalLM",
589
+ "DeepseekV32ForCausalLM",
590
"Glm4MoeForCausalLM",
591
):
592
E = config.n_routed_experts