Saving model results to a pandas dataframe?

Hi all, I am experimenting with some callbacks at the moment and I was wondering if there was a way to save the best models results into a pandas dataframe declared out of the function that I pass in? I am currently using the save_model callback to get the best value, but I want that value stored into a dataframe for comparison. Thank you!

1 Like

Hi everybody,
I have a similar problem which I couldn’t find a solution for.

I train a NN with train_one_cycle() and now I need to store the written accuracy in a variable. Any ideas how this might work?

Here’s a blogpost that I found this on:

In the notebook on Hyperparameter Optimization they succeed in doing this by doing float(learn.validate([1]), which I succeeded in finding as well. This extracts the most recent accuracy as a float variable we can then load into pandas or what have you. Hopefully this may help some of you.