Timothee Mickus
Timothee Mickus
@shaoxiongji this could impact docs.
smoketested with hydra config on 1 GPU during test. Seems to be working properly.
utils for loading an existing checkpoint from mammoth: https://github.com/Helsinki-NLP/mammoth/blob/main/onmt/model_builder.py#L179C9 all CLI/ config file options: listed in https://github.com/Helsinki-NLP/mammoth/blob/main/onmt/opts.py (in particular, see `train_opts` for structural options, and `build_bilingual_model` )
indices for special tokens are here: https://github.com/Helsinki-NLP/mammoth/blob/main/onmt/inputters/vocab.py#L10
[The `backports.zoneinfo` dependency should be conditioned on python
Embarks partial fix for #7
@Waino do we merge?
Hi! Apologies for the late answer. Other input types are currently not implemented, although we've had this request more than once. We don't really have enough hands to look into...
Yes, early stopping should be supported out of the box. Assuming you want to evaluate every 10k steps, and stop training if it no longer improves after 5 evaluation loops,...
No worries! don't hesitate to share your code if you want us to include it in the library, we would welcome a pull request if you have something working.