We need to write a separate inference script that evaluates a fine-tuned model.
Currently, we do training and evaluation with the train_and_evaluate function in finetune.py, but there are some datasets that have external or multiple test sets. Therefore, it will be more useful to be able to make inference externally. In addition, it will make it easier to prepare and test the postprocess and compute_metrics functions for each dataset.
We need to write a separate inference script that evaluates a fine-tuned model.
Currently, we do training and evaluation with the
train_and_evaluate
function infinetune.py
, but there are some datasets that have external or multiple test sets. Therefore, it will be more useful to be able to make inference externally. In addition, it will make it easier to prepare and test the postprocess and compute_metrics functions for each dataset.