Maksym Del

Fix transformers vocab not being saved to files

1

Fixes https://github.com/allenai/allennlp/issues/3097. Here we add a method to vocab to populate its namespace from transformers encoding (which is fetched in indexer) with an option to resave vocab to files (which...

Seems like the tokens get wrapped in a TextField twice. Interesting that the model's code doesn't break.

I am an extremely regular AWS folk, but it looks it is possible to retain the data after the spot instance terminates: https://aws.amazon.com/premiumsupport/knowledge-center/ami-preserve-ebs-spot/ Could this feature be added to the...

XLM-R Hyperparameters

2

Hi @JunjieHu! I am trying to reproduce results for XLM-R. The paper suggests that lr=3e-5 and effective bs=16 should be used for XLM-R. It would be very helpful if you...

Add PrefixSuffix multiseq mappers to prepend/append tokens

3

Hi! With the introduction of Smashed, munging datasets of long documents is going to be a lot more fun ) This draft PR is simply to explore the idea below....

These are unique mappers since they read/write data from/to disc, so I made a new file for them. I think they can provide three primary benefits to the users: 1)...