Update modeling_deepseek_v32.py

modeling_deepseek_v32.py

@@ -663,7 +663,7 @@ class DeepseekV32DecoderLayer(GradientCheckpointingLayer):
         self.hidden_size = config.hidden_size
         self.self_attn = DeepseekV32Attention(config, layer_idx)
 
-        if config.mlp_layer_types[layer_idx] == "sparse":
+        if config.first_k_dense_replace < layer_idx:
             self.mlp = DeepseekV32MoE(config)
         else:
             self.mlp = DeepseekV32MLP(config)