Support Pangu Pro MoE model (#1204)

### What this PR does / why we need it? Support Pangu Pro MoE model (https://arxiv.org/abs/2505.21411) ### Does this PR introduce _any_ user-facing change? Yes, new model supported ### How was this patch tested? Test locally --------- Signed-off-by: angazenn <zengyanjia@huawei.com> Co-authored-by: angazenn <zengyanjia@huawei.com>
2025-06-20 23:59:59 +08:00
parent 00ae250f3c
commit 2f1266d451
4 changed files with 647 additions and 1 deletions
--- a/vllm_ascend/ops/common_fused_moe.py
+++ b/vllm_ascend/ops/common_fused_moe.py
@@ -43,6 +43,7 @@ def forward_oot(
    activation: str = "silu",
 ) -> torch.Tensor:
    topk_weights, topk_ids = select_experts(
+        global_num_experts=global_num_experts,
        hidden_states=x,
        router_logits=router_logits,
        top_k=top_k,