This summer I had the pleasure to (co-)mentor at Google Summer of Code for TensorFlow. I was responsible for mentoring two students: Aditya and Vasudev.
They implemented and trained SOTA models for image recognition and speech and published them on TensorFlow Hub. We now have RegNets (for vision) and wav2vec2 (for speech) with clean training code and pre-trained models open-sourced all in TensorFlow and Keras. Cc: @fchollet.
I had a chance to author this blog post with another fellow mentor Morgan Roff of Google. If you’re a university student and would want to contribute SoTA models in an open-source spirit, do check it out.