What is the format of the training data used for fine-tuning, or an example?
A format similar to the inference is used. prefix + source and target as labels.
· Sign up or log in to comment