Post date: Mar 10, 2018 12:31:33 PM
Topic: Recurrent neural networks
Video: Lec 10
Reading: DL Book Chap 10, Knet draft chapter, Tutorials: Karpathy 2015, Olah 2015, Britz 2015, Hinton 2012, Advanced: Olah and Carter 2016, Graves 2012, Other references
Notebooks: RNN tutorial, Sentiment analysis, RNN language model, Character based language model
Midterm: Thursday March 29 in class.
Project: Baseline model due March 23. Implement a baseline model that generates random outputs. Your baseline model should be able to minibatch and read your train/dev/test data, calculate the average train/dev/test loss and task-specific evaluation metric (e.g. accuracy) correctly. Update your Google Doc, Sheet and GitHub repo to complete the assignment.
Late completion will be graded based on the following formula: (1/2)+(1/2)^(d/3+1) where d is the number of days late.