Closed JackKelly closed 2 years ago
Start simple: just do step 2. Then sample freely across data in men (as we currently do)
Just done item 2 on the list above
Seems to reduce validation loss a bit. I'm going to count this as "done", even though I've only done item 2 on the list above.
Start simple: just do step 2. Then sample freely across data in men (as we currently do)