add NEFTune optimization

Former-commit-id: 603e0298af64116ac07130fe6661a9ba823c186c

src/llmtuner/hparams/finetuning_args.py

@@ -75,6 +75,10 @@ class FinetuningArguments:
         default=0.1,
         metadata={"help": "The beta parameter for the DPO loss."}
     )
+    neftune_noise_alpha: Optional[float] = field(
+        default=None,
+        metadata={"help": "The alpha parameter for the NEFTune noise. By setting this the NEFTune optimization will be activated."}
+    )
 
     def __post_init__(self):
         if isinstance(self.lora_target, str): # support custom target modules/layers of LoRA