Redo BERT fine-tuning (and downstream) experiments with iso reg_type and reg_str NOT 0

mmatena / m251

0 stars 0 forks source link

Open mmatena opened 3 years ago

mmatena commented 3 years ago

I had a bug where the reg strength would be 1 if the reg strength was not 0.

mmatena commented 3 years ago

I also should fix the switch up between the experiment and run uuids in FisherIsoParams before re-running.

mmatena commented 3 years ago

Also maybe add some more GLUE tasks. See which ones I have on GCP.