spro/practical-pytorch

Covers character-level and sequence-to-sequence RNN architectures for NLP tasks including text classification, generation, and machine translation with attention mechanisms. Implements practical applications like name classification, Shakespeare text generation, and neural machine translation using PyTorch's autograd and embedding layers. Jupyter notebooks demonstrate end-to-end workflows with internet-sourced datasets and GloVe word vectors.

4,543 stars. No commits in the last 6 months.