you are a developing a large language model using a GPT $3$ and want to apply few short learning technique you have a limited data set for a specific task can you want the model to generalized by which of the following approach would be the most effective train the model on the entire data set then find unit on a small subset of a task specific data provide a few example of the task in the input and let the model generate a response without any gradient updates of Mind data set for training and apply Radiant updates only on the task specific data train the model on the last data set and apply few short learning technique on the entire data set or use of small data set for training and apply Radiant updates only on the task specific data