Brad Jascob

The annotator script (see readme 3. step) creates the tokens, lemmas,... tags. This uses the Stanford NLP system. Using NLTK to annotate will likely give you less than optimal results...

For some reason, thre's an if statment in `stog/commands/predict.py::_predict_instances()` (about line 141) that sends the code on a different path if your `batch_data` size is 1 (ie.. only one sentence)....

I'm interested in this as well but I haven't had time to work on it. The original paper "retrofitted" T5 by adding additional cross-attentions between the pretrained model and the...

No. The transformer probably learned a word like this during pre-training and now it's outputting the spelling that it learned previously. Generally, we have no control over this.

Closing. No activity on issue for 2 months.

I would recommend preprocessing the LDC data with the script [10_CollateData.py](https://github.com/bjascob/amrlib/blob/master/scripts/20_Assemble_LDC2020T02/10_CollateData.py). This script also combines all files of a specific type (ie.. test) into a single file. If you don't...

This is actually flaw in the LDC data. I reported the error to them but I'm not sure when/if it'll ever get fixed. I hate to just run all incoming...

See #19. This points to a few places in the code to change if you want to do this yourself or there's another suggested work-around. At some point in the...

The work-around should only be a few lines of code to implement. It might not be the "right" solution but it should work fine. Just lower-case match the output aligned...

If you want to drop a tar-ball in here with your code, I'll take a look at some point in the future.