Metrics for Fine-Tuned binary Classificator


I’m wondering if I can get the precision, recall and f1-score for the positive and negative classes of my fine-tuned binary Classificator model? I don’t understand why I get them only for the positive class. It would also be Ok if I just get the TP, FP, FN and FP so that I can calculate them on my own.

I don’t really want to have all the validation data predicted by the model again. Since the model has already created the metrics for the positive class, it should also be able to create the metrics for the other class, right?

For the creation of the Fine-Tuned model I used the following command:

openai api fine_tunes.create -t "data_finetuning_prepared_train.jsonl" -v "data_finetuning_prepared_valid.jsonl" -m ada --compute_classification_metrics --classification_positive_class " None"

I hope somebody can help me :slight_smile: