Skip to content
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.

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

add orthogonal softmax configs #87

Merged
merged 2 commits into from
Jan 12, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view

Large diffs are not rendered by default.

Large diffs are not rendered by default.

8 changes: 8 additions & 0 deletions 2024-orthogonal-softmax/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -4,4 +4,12 @@ Efficient Supernet Training with Orthogonal Softmax for Scalable ASR Model Compr

We use [RETURNN](https://github.com/rwth-i6/returnn) for training and our setups are based on [Sisyphus](https://github.com/rwth-i6/sisyphus).

We use models parts from [i6-models](https://github.com/rwth-i6/i6_models/tree/jing-dynamic-encoder-size)

### subnet selection with component-wise criterion

ConformerCTCModel, ConformerCTCConfig and train_step in returnn config is defined in [here](https://github.com/rwth-i6/i6_experiments/blob/main/users/jxu/experiments/ctc/tedlium2/pytorch_networks/dynamic_encoder_size/orthogonal_softmax/joint_train_conformer_orthogonal_softmax_component_wise.py)

### subnet selection with layer-wise criterion

ConformerCTCModel, ConformerCTCConfig and train_step in returnn config is defined in [here](https://github.com/rwth-i6/i6_experiments/blob/main/users/jxu/experiments/ctc/tedlium2/pytorch_networks/dynamic_encoder_size/orthogonal_softmax/joint_train_conformer_orthogonal_softmax_layer_wise.py)
Loading
Loading