Update src/llama_recipes/finetuning.py

src/llama_recipes/finetuning.py

@@ -107,7 +107,7 @@ def main(**kwargs):
         use_double_quant=True,
         quant_storage=torch.bfloat16
     ) if train_config.quantization else None
-    bnb_config = quant_config.create_bnb_config()
+    bnb_config = quant_config.create_bnb_config() if quant_config else None
 
     # Load the pre-trained model and setup its configuration
     use_cache = False if train_config.enable_fsdp else None