Learn how to use PyTorch to solve some common NLP problems with deep learning. View these notebooks on nbviewer.

• 1_BoW_text_classification.ipynb: Train a bag-of-words model to predict the sentiment of IMDB reviews
• 2_embeddings.ipynb: Play around with different pretrained word embeddings
• 3_rnn_text_classification.ipynb: Train an RNN to predict the sentiment of IMDB movie reviews
• 4_character_text_generation.ipynb: Train a character-level RNN language model to generate weight loss articles
• 5_seq2seq_attention_translation.ipynb: Train an RNN-based Seq2Seq model with attention to translate from English to French
• 6_transformer_translation.ipynb: Train a pure self-attention based transformer Seq2Seq model to translate from English to French
• 7_gpt2_finetuned_text_generation.ipynb: Fine-tune the pretrained (small) GPT-2 model to generate weight loss articles

Tutorials

Make sure you have a Google account and visit Google Colab. You should see a list of notebooks pop up:

If you have trouble with that, you can also save the notebook you want to run from this repo to your local filesystem, and then upload it to Google Colab with File -> Open Notebook -> Upload.

You can run cells with <SHIFT> + <ENTER>.

If you find that you are missing a necessary package, you can prepend ! to a bash command. For example, to install googledrivedownloader, you would run in a cell:

!pip install googledrivedownloader

To use a GPU (for free!), select from the top menu from Colab Runtime -> Change Runtime Type -> Hardware Accelerator -> GPU. Pay attention to how much memory the GPU is currently using by clicking Runtime -> Manage Sessions.

Feel free to submit a PR for cleanups, error-fixing, or adding new (relevant) content!