Assignment 3 Checklist

Thresholds

[ ] 3 or 5 epoch burn-in [ ] Dev loss has either gone down less than some epsilon or has increased relative to last epoch

Experiments with Czech

[] Average and Max embedding combinations [] Try different thresholds/burn-in combinations (to be listed later)

Autoencoder

[] Change embedding combination to retrieve all sets that it needs to combine [] Pass sets into separate module which contains RNN [] Split sets into train/dev [] Train RNN [] Compute embeddings once it gets below certain loss on dev [] Retain RNN, but discard optimizer. [] Repeat on future runs - Need to decide on the following:

• Should we retrain or should we only evaluate the trained RNN on the new examples?
• Should we retain the old training data (previous combinations), or only use the new training data?