Closed danich1 closed 6 years ago
Do we keep track of training and testing performance (sentence-level) after each epoch? It would be nice to visualize how much each epoch changes performance.
Their script outputs the loss for each epoch. I don't have the values saved for this run, because of gpu cluster issues. Following the downstream experiments, I'll be saving that information as well.
Great. @danich1 will let you merge this when you're ready.
Quick pr for adding updated LSTM results. Ran for 22 epochs and slowly see an increase in test predictions. The other files edited are small bug fixes I encountered while generating the LSTM results.