Currently, the function tsne_model in tsne_model_embeddings.py can only deal with PyTorch models. However, the function plot_tsne still works with the outputs of linear models when predictions are generated via the predict_proba method.
When writing the Jupyter notebook #14 for reproducing our result, keep this in mind and add a code block that deals with generating the t-SNE embeddings for the linear model.
Currently, the function
tsne_modelintsne_model_embeddings.pycan only deal with PyTorch models. However, the functionplot_tsnestill works with the outputs of linear models when predictions are generated via thepredict_probamethod.When writing the Jupyter notebook #14 for reproducing our result, keep this in mind and add a code block that deals with generating the t-SNE embeddings for the linear model.