CUNY-CL/yoyodyne Issues
TQDM Error with multi GPU Transducer
Updated 1Hard Monotonic Transducer
Closed 3Automatic batch sizing
Updated 10SER on GPU
Closed 3Add more evaluation metrics
Closed 1Bump maxwell version to >=0.2.3
Closed 1W&B sweeps giving OOM
Closed 10GRU support
UpdatedShared embeddings
Updated 4Torch and Lightning 2.0.0
Updated 6Training UNK tokens
Updated 14Subword tokenization
Updated 4Model support: GPT
Updated 6Implement Pytorch Metrics
Updated 4Friendly predictor interface
Updated 1Add self attention encoder
Updated 1Switch to CrossEntropyLoss?
Closed 5Tied vocabulary flag: no-op?
Closed 3Python interface for loading
Closed 8Validation Accuracy Aggregation
Closed 3Number of layers have to match
Closed 3Feature-invariant LSTM
Closed 4Informative model logging
Closed 1Feature indexing
Closed 4ByT5 encoder
UpdatedBERT Pretraining
Closed 5Default Scripts
Closed 2Multi-layer LSTMs broken
Closed 2