Hi Open AI community
I’m looking for help regarding the fine-tuning of davinci with more than 10k data.
I have followed the data preparation and the splitting of train and validation dataset.
A prompt example
{"prompt":"musty smell in corridors ->","completion":" 66"}
The data has 220 categories.
When I create the fine tune model in a jupyter notebook using
!openai api fine_tunes.create -t "train.jsonl" -v "valid.jsonl" --compute_classification_metrics --classification_n_classes "220" -m davinci
It keeps loading with nothing happening.
I have also tried curie and ada, same observation.
I am on a paid plan, and every step is producing the expected outcome up to the creation of the fine tune model.
Any idea what I’m missing? What it is slow/not working?