Update qwen3_moe.py

vllm/model_executor/models/qwen3_moe.py

@@ -332,9 +332,9 @@ class Qwen3MoeModel(nn.Module):
         self.padding_idx = config.pad_token_id
         self.vocab_size = config.vocab_size
         self.config = config
-        if self.config.quantization_config["bits"] == 4:
-            os.environ['LLAMA_NN'] = '0'
-            os.environ['LM_NN'] = '0'  
+        # if self.config.quantization_config["bits"] == 4:
+            # os.environ['LLAMA_NN'] = '0'
+            # os.environ['LM_NN'] = '0'  
         self.embed_tokens = VocabParallelEmbedding(
             config.vocab_size,
             config.hidden_size,