batch_sentences() got an unexpected keyword argument 'max_length'

[thomas-davis]

Bilm Batcher.batch_sentences (here: https://github.com/allenai/bilm-tf/blob/master/bilm/data.py) doesn't accept a max_length argument. I can't find a version in their github history that does. Any work around? Can I safely remove the max_length argument.
Thanks!

[haoyangz]

Have you run python update_blim.py as mentioned in the README?

[thomas-davis]

woops, thank you, resolved.

[thomas-davis]

Also can I ask, how would one convert column 1 and 2 of the prediction output into nanomolar affinities (as I think is done in fig 4 of your Bioinformatics paper). Is this conversion possible?
Thanks again.

[haoyangz]

@thomas-davis conversion between nanomolar and the normalized scaled is described in the second paragraph in section 2.1 of the paper. For fig 4, I believe the affinity mean prediction (first column of the output) was used to convert to nanomolar.

[thomas-davis]

Thank you again! one last question-- I'd like to run the affinity only or embedding only model for a dataset, and wasn't sure how to go about it.