Book: "Neural Machine Translation"

Philipp Koehn June 2020 Hardcover, 394 pages Publisher: Cambridge University Press ISBN-10: 1108497322 ISBN-13: 978-1108497329

Available from amazon.com.

Content

I Introduction

• Chapter 1: The Translation Problem
• Chapter 2: Uses of Machine Translation
• Chapter 3: History
• Chapter 4: Evaluation

• Chapter 5: Neural Networks
• Chapter 6: Computation Graphs
• Chapter 7: Neural Language Models
• Chapter 8: Neural Translation Models
• Chapter 9: Decoding

• Chapter 10: Machine Learning Tricks
• Chapter 11: Alternate Architectures
• Chapter 12: Revisiting Words
• Chapter 13: Adaptation
• Chapter 14: Beyond Parallel Corpora
• Chapter 15: Linguistic Structure
• Chapter 16: Current Challenges
• Chapter 17: Analysis and Visualization

Slide sets that cover the book closely will be developed in Fall 2020 for the JHU class on machine translation. The most recent edition of the class has material posted at mt-class.org.

Code

• Chapter 5: a basic neural network for xor nn.py
• Chapter 6: xor example with pytorch torch_basic.py (Sec. 6.3.1), torch_batch.py, torch_loop.py (Sec. 6.3.2), torch_loop_batch_update.py (Sec. 6.3.3), torch_module.py (Sec. 6.3.4)
• Chapter 7: recurrent neural network rnn.py
• Chapter 8: neural machine translation nmt.py, Tanzil.20k.de-en.de, Tanzil.20k.de-en.en
• Chapter 9: decoder decoder.py