Learner previous epochs/runs/stats

I have a Jupyter notebook running on beefy Linux box I built that I can access over the LAN, so I can start a fitting a model overnight, and I don’t have to worry about getting Optimus graphics working with Jupyter on my laptop(nightmare!)

So if I start running say:

learn.fit_one_cycle(25, lr)

And close the browser or laptop and come back the next day, if I refresh the page I see something like: https://i.imgur.com/8e9TqrF.png

Is there a way to view the previous epochs with all the loss and accuracy for each run? Something like

learn.previous_cycle_stats()

It would print out these stats:

Hmm half answer:

learn.recorder.metrics_names, learn.recorder.metrics

which returns:

I don’t get a nice pandas dataframe like table though with the extra stats.