Clear output

microsoft · Dec 16, 2023 · 880417e · 880417e
1 parent c1e90a3
commit 880417e
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/deepspeed/inference/v2/model_implementations/common_parameters/moe_parameters.py b/deepspeed/inference/v2/model_implementations/common_parameters/moe_parameters.py
@@ -75,4 +75,4 @@ def finalize(self) -> torch.Tensor:
             transposed_experts.append(fused_expert)
 
         stacked_experts = torch.stack(transposed_experts, dim=0)
-        return self.inference_model.transform_moe_mlp_2_param(stacked_experts)
+        return self.inference_model.transform_moe_mlp_1_param(stacked_experts)
diff --git a/deepspeed/inference/v2/model_implementations/mixtral/container.py b/deepspeed/inference/v2/model_implementations/mixtral/container.py
@@ -36,7 +36,7 @@ class MixtralTransformerContainer(LayerContainer):
 class MixtralNonTransformerContainer(LayerContainer):
 
     word_emb: EmbeddingParameter
-    word_unembed: EmbeddingParameter
+    word_unembed: UnembedParameter
     final_norm: NormParameter
 
     PARAM_MAPPING = {