Wine Predictions Based on Sommelier Reviews (NLP with a ULMFit Network)
Image source: ULMFiT paper by Jeremy Howard and Sebastian Ruder
Using PyTorch and Deep Learning applications in NLP, this project implements a ULMFit network to incorporate transfer learning in an NLP framework to predict the type of wine from a Sommelier review.
The RNN model utilizes transfer learning from the WT103 pretrained language model and was then further trained on a dataset scraped from WineEnthusiast on November 22nd, 2017.
Daniel Diamond
data watches music travel