This PR is uploading my previous experiment where we tested to see if label functions could transfer between relationship types. Majority of this PR contains files that I added to run this experiment. To save time only concern yourself with the transfer_learning.ipynb file. Note: reason for erratic performance when training the models is that I am predicting only 50 sentences. For future reference more data means more stable results.
This PR is uploading my previous experiment where we tested to see if label functions could transfer between relationship types. Majority of this PR contains files that I added to run this experiment. To save time only concern yourself with the transfer_learning.ipynb file. Note: reason for erratic performance when training the models is that I am predicting only 50 sentences. For future reference more data means more stable results.