Added
- PyTorch version Transformer & SpeechTransformer model.
- Audio extraction for CommonVoice/IWSLT.
- Data sampler and dataset for multilingual machine translation
- Mixed training dataset with data sampler.
- Multilingual Translation task
- Instruction for
- training transformer models on WMT14 EN->DE
- weight pruning
- quantization aware training for transformer model
Fixed
- Compat with TensorFlow v2.4