Text-to-Text Translation

The objective is to extensively study the available deep learning solutions for the text to text translation problem using the French to English text dataset from hugging face.

You can find the detailed analysis in the report.pdf, and the code in the notebooks directory. Also, check out the presentation for a quick overview of the project.

Models

The following models are used for the text-to-text translation problem:

1. seq2seq
2. seq2seq with Bidirectional and Embedding Layers
3. seq2seq encoder-decoder with attention mechanism
4. seq2seq encoder-decoder embedding bidirectional with attention mechanism
5. Transformer (T-5 small)
6. Torch Transformer (trained on the dataset)