Neural Nets for NLP 2021 - Language Modeling, Efficiency/Training Tricks

Neural Nets for NLP 2021 - Language Modeling, Efficiency/Training Tricks

Graham Neubig via YouTube Direct link

Intro

1 of 29

1 of 29

Intro

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Neural Nets for NLP 2021 - Language Modeling, Efficiency/Training Tricks

Automatically move to the next video in the Classroom when playback concludes

  1. 1 Intro
  2. 2 Language Modeling: Calculating the Probability of a Sentence
  3. 3 Count-based Language Models
  4. 4 A Refresher on Evaluation
  5. 5 Problems and Solutions?
  6. 6 An Alternative: Featurized Models
  7. 7 A Computation Graph View
  8. 8 A Note: "Lookup"
  9. 9 Training a Model
  10. 10 Parameter Update
  11. 11 Unknown Words
  12. 12 Evaluation and Vocabulary
  13. 13 Linear Models can't Learn Feature Combinations
  14. 14 Neural Language Models . (See Bengio et al. 2004)
  15. 15 Tying Input/Output Embeddings
  16. 16 Standard SGD
  17. 17 SGD With Momentum
  18. 18 Adagrad
  19. 19 Adam
  20. 20 Shuffling the Training Data
  21. 21 Neural nets have lots of parameters, and are prone to overfitting
  22. 22 Efficiency Tricks: Mini-batching
  23. 23 Minibatching
  24. 24 Manual Mini-batching
  25. 25 Mini-batched Code Example
  26. 26 Automatic Mini-batching!
  27. 27 Code-level Optimization . eg. TorchScript provides a restricted representation of a PyTorch module that can be run efficiently in C++
  28. 28 Regularizing and Optimizing LSTM Language Models (Merity et al. 2017)
  29. 29 In-class Discussion

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.