Simplifying ML Programming
Soumith explains the cognitive overhead of symbolic programming in Tensorflow compared to the simplicity of Pytorch's Python-based approach. He discusses the challenges in building an ML compiler mode that seamlessly handles program compilation without burdening the user.In this clip
From this podcast
Gradient Dissent - A Machine Learning Podcast
Exploring PyTorch and Open-Source Communities: Interview with Soumith Chintala of Meta & PyTorch
Related Questions