Embeddings-for-trees | 1.2.0

• Build vocabulary based on counts of tokens
• Detailed wandb configuration
• Return attention weights from forward
• Upgrade dependencies to Torch, DGL, Lightning, etc
• Fix bugs

Fix reseting metrics

Call metric reset on each epoch end

Improved uploading model checkpoints

1.0.3

Update requirements, new logic for uploading weights

Add all modules into the package

1.0.2

Add init files for correct package resolving

First release of embeddings-for-trees

The first release of embeddings for trees package.
This release includes TreeLSTM PyTorch modules and lightning module to train and evaluate it.