Removing bitsandbytes quantization to 8 bits

model_translation.py

@@ -209,8 +209,8 @@ class ModelM2M100(metaclass=Singleton):
             self._model_name,
             device_map="auto",
             torch_dtype=torch.float16,
-            low_cpu_mem_usage=True,
-            quantization_config=quantization_config
+            low_cpu_mem_usage=True
+            #quantization_config=quantization_config
         )
         self._model = torch.compile(self._model)