My project on Word Vectors
An implementation of the skip gram model using Tensorflow
The source vectors were first learned without the regularized loss function using the same code. Source domain corpus can be found here (from Tensorflow's implementation)- http://mattmahoney.net/dc/
The target domain corpus was then used to calculate the regularization parameters after basic text preprocessing. The regularization term was added to the loss function. Source domain corpus (Amazon Fine Food Reviews)- https://www.kaggle.com/snap/amazon-fine-food-reviews/data
Current issue- Looking for efficient methods for multiplying tensors