update test_moe.py

tests/kernels/test_moe.py

@@ -171,8 +171,8 @@ def test_mixtral_moe(dtype: torch.dtype):
     for i in range(config.num_local_experts):
         weights = (hf_moe.experts[i].w1.weight.data,
                    hf_moe.experts[i].w3.weight.data)
-        vllm_moe.experts.w13_weight[i][:] = torch.cat(weights, dim=0)
-        vllm_moe.experts.w2_weight[i][:] = hf_moe.experts[i].w2.weight.data
+        vllm_moe.experts.w13_weight[i][:] = (torch.cat(weights, dim=0)).T
+        vllm_moe.experts.w2_weight[i][:] = (hf_moe.experts[i].w2.weight.data).T
 
     # Generate input batch of dimensions [batch_size, seq_len, hidden_dim]
     hf_inputs = torch.randn((1, 64, config.hidden_size)).to(dtype).to("cuda")