New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

It should be possible to use different neural backends #550

Open

kwalcock opened this issue Sep 13, 2021 · 3 comments

Member

kwalcock commented Sep 13, 2021

No description provided.

Member Author

kwalcock commented Sep 13, 2021 •

edited

Training data is on the servers at /data/nlp/corpora/processors-dynet
Resources for training named mtl-en...
Config library

Member Author

kwalcock commented Sep 13, 2021 •

edited

Example TorchScript Code
Branch supporting multiple backends
Glove embeddings - unzip jar for txt file

Member Author

kwalcock commented Sep 13, 2021 •

edited

Start with NER because simplest
Run training in Scala and record performance numbers (see those shell scripts)
Do in Python and compare performance

Parse config files
Implement TaskManager (Shard, Task)
Change AnnotatedSentence to also receive a sequence of Glove vectors
Implement Meta for NER

src/main/python
metal/pytorch

kwalcock mentioned this issue

Make room for other backend (NN) implementations #549

Draft

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment