CS 20SI: TensorFlow for Deep Learning Research

Code examples, assignments and solutions for the course CS 20SI: TensorFlow for Deep Learning Research (winter 2017).

Detailed syllabus and lecture notes can be found at http://cs20si.stanford.edu

Models include:

Examples:

• Linear Regression with Chicago's Fire-Theft dataset
• Logistic Regression with MNIST
• Word2vec skip-gram model with NCE loss
• Convnets with MNIST
• Autoencoder (by Nishith Khandwala)
• Deepdream (by Jon Shlens)
• Character-level language modeling

Assignments:

• Style Transfer
• Chatbot using sequence to sequence with attention

Misc

• Examples on how to use data readers, TFRecord
• Embedding visualization with TensorBoard
• Usage of summary ops
• Exercises to be familiar with other special TensorFlow ops
• Demonstration of the danger of lazy loading
• Convolutional GRU (CRGU) (by Lukasz Kaiser)