texar-pytorch
texar-pytorch copied to clipboard
asyml
Integrating the Best of TF into PyTorch, for Machine Learning, Natural Language Processing, and Text Generation. This is part of the CASL project: http://casl-project.ai/
"What worries me is that our implementation of modules based on pre-trained stuff is a bit too repetitive, so that a seemingly small change would require modifying a bunch of...
When subclassing existing modules, a common use case I've encountered is to automatically fill certain arguments of the super class constructor based on HParams values. By design, HParams are realized...
Configurations in examples (e.g. [transformer/config_model.py](https://github.com/asyml/texar-pytorch/blob/master/examples/transformer/config_model.py)) is written in a separate file with config values as global variables. To be able to use these config files, we *import* them with `importlib`,...
Keras-tuner (https://github.com/keras-team/keras-tuner) is a hyperparameter tuner for `tf.keras`. We like to take insights from this framework. In particular, we like to design the following interface in our framework 1) Develop...
Currently, we set `vocab.bos_token` and `vocab.eos_token` to their default values (`` and `` respectively) when `bos_token` and `eos_token` in hparams are set to empty string. This creates inconsistencies between the...
As pointed out in #110, it would be useful to record hidden states in return values. A problem with the current interface is we only have type annotation for `State`,...