NLG Head to Toe

This is a gentle 3 hours tutorial on Natural language Generation using Neural Networks, the tutorial starts from the basics and assumes very little background on NLP and Machine Learning.

Content

  1. Language Modeling

  2. Recurrent Neural Networks

  3. NLG Inference (Top-k and Nucleus Sampling, Greedy Decoding)

  4. Language Model Evaluation ( Perplexity and Cross-Entropy)

  5. Conditional Language Models

  6. Attention is all you Need (Transformers)

  7. Large Language Models and their dangers

Tutorial Instances

  • 26 Feb. 2021 - CentraleSupélec & ESSEC, Paris, France

  • 3 Dec 2021. - Hanoi University, Hà Nội, Vietnam

7XjH1Zk.png
1sXTE0n.png

If the slides below don’t load , you can find the slides here