guillaumekln

4 ms·

1. 1.▲
   by guillaumekln 3y ago

   The original Whisper implementation from OpenAI uses the PyTorch deep learning framework. On the other hand, faster-whisper is implemented using CTranslate2 [1] which is a custom inference engine for Transformer models. So basically it is r
2. 2.▲
   OpenNMT-tf: A modular TensorFlow implementation for neural sequence learning(github.com)

   3 points by guillaumekln 9y ago|0 comments