I have added the confidence score functionality. I simply added a Softmax activation to the last layer of the Transformer decoder and now use the values that are generated for a token-level confidence estimation.
TODO (I would do it myself, but I don't think I have access to the Zenodo repository):
Repack the model (code is already updated)
Upload the updated repacked model to Zenodo
Update the Zenodo link in decimer.py
I'm not sure if a draft PR triggers Github actions; if it does, I apologise for the failed test notifications. The tests are only going to pass when the updated repacked model is included.
I have added the confidence score functionality. I simply added a Softmax activation to the last layer of the Transformer decoder and now use the values that are generated for a token-level confidence estimation.
TODO (I would do it myself, but I don't think I have access to the Zenodo repository):
decimer.py
I'm not sure if a draft PR triggers Github actions; if it does, I apologise for the failed test notifications. The tests are only going to pass when the updated repacked model is included.