Esquema per temes

• Select topic General

  General

  Contreu-ho tot Expandeix-ho tot

  Target audience: The course is recommended for master students of all CS-oriented programs (Master Informatik, Master eXtended AI, Master Business Informatics). Prior knowledge of core NLP and machine learning concepts is desirable, albeit not mandatory. 

  Credits: 5 ECTS

  Lectures: Wed, 10.15-11.45 in Übungsraum I (ÜR I, Informatikgebäude)

  Exercises: Fri, 14.15-15.45 in Übungsraum I (ÜR I, Informatikgebäude)

  Examination: written or oral exam (depends on the number of students, it will be fixed and announced after the first two weeks of lectures)

  
  

  • Select activity Announcements

    

    Announcements Fòrum
  • Select activity Discussion forum

    

    Discussion forum Fòrum
• Select topic Lectures

  Lectures
  • Select activity Lecture 1: Language Diversity (+Course Organization)

    

    Lecture 1: Language Diversity (+Course Organization) Fitxer
  • Select activity Lecture 2: Neural Language Modeling & Tokenization

    

    Lecture 2: Neural Language Modeling & Tokenization Fitxer
  • Select activity Lecture 3: Training Deep NNs

    

    Lecture 3: Training Deep NNs Fitxer
  • Select activity Lecture 4: Transformer

    

    Lecture 4: Transformer Fitxer
  • Select activity Lecture 5: Cross-Lingual Word Embeddings

    

    Lecture 5: Cross-Lingual Word Embeddings Fitxer
  • Select activity Lecture 6: Cross-Lingual Transfer

    

    Lecture 6: Cross-Lingual Transfer Fitxer
  • Select activity Lecture 7: Curse of Multilinguality

    

    Lecture 7: Curse of Multilinguality Fitxer
  • Select activity Lecture 8: Word Alignment and Label Projection

    

    Lecture 8: Word Alignment and Label Projection Fitxer
  • Select activity Lecture 8 (Word alignment): Recording

    

    Lecture 8 (Word alignment): Recording URL
  • Select activity Lecture 9: Text Generation (with NMT focus)

    

    Lecture 9: Text Generation (with NMT focus) Fitxer
  • Select activity Lecture 10: (Multilingual) Sentence Encoders

    

    Lecture 10: (Multilingual) Sentence Encoders Fitxer
• Select topic Project

  Project
  • Select activity Submission project presentation and code

    

    Submission project presentation and code Tasca
• Select topic Exercises

  Exercises
  • Select activity Exercise 01

    

    Exercise 01 Fitxer
  • Select activity Exercise 02

    

    Exercise 02 Fitxer
  • Select activity Exercise 03

    

    Exercise 03 Fitxer
  • Select activity Exercise 04

    

    Exercise 04 Fitxer
  • Select activity Projects

    

    Projects Fitxer
  • Select activity Exercise 05 - Parameter-Efficient Fine-Tuning

    

    Exercise 05 - Parameter-Efficient Fine-Tuning Tasca
  • Select activity Exercise 05 - Parameter-Efficient Fine-Tuning - Solutions

    

    Exercise 05 - Parameter-Efficient Fine-Tuning - Solutions Fitxer
  • Select activity Exercise 06 - Token-Level Transfer

    

    Exercise 06 - Token-Level Transfer Tasca
  • Select activity Exercise 06 - Token-level Transfer - Solutions

    

    Exercise 06 - Token-level Transfer - Solutions Fitxer
  • Select activity Exercise 07 - Neural Machine Translation

    

    Exercise 07 - Neural Machine Translation Tasca
  • Select activity Exercise 07 - Neural Machine Translation - Solutions

    

    Exercise 07 - Neural Machine Translation - Solutions Fitxer
  • Select activity Exercise 08 - Sentence Representation Learning

    

    Exercise 08 - Sentence Representation Learning Tasca
  • Select activity Exercise 08 - Sentence Representation Learning - Solutions

    

    Exercise 08 - Sentence Representation Learning - Solutions Fitxer
• Select topic Course content

  Course content

  This is planned/tentative content. As such it is still subject to change, that is, modifications both with respect to the topics and order of lectures are possible. 

  Date	Lecture	Blocks
  26.4.	L1: Languages of the world & Linguistic Universals; Course organization	Introduction
  10.5.	L2: Language modeling, word embedding models, vocabulary & tokenization	Block I: Fundamentals
  17.5.	L3: Training Deep LMs — Gradient Descent, Backprop, Batching, Dropout…	Block I: Fundamentals
  24.5.	L4: Transformer Almighty & Pretraining Language Models	Block I: Fundamentals
  31.5.	L5: Multilingual Word Embedding Spaces	Block II: Multilinguality
  7.6.	L6: Multilingual LMs and CL Transfer, Multilingual Evaluation	Block II: Multilinguality
  14.6.	L7: Curse of Multilinguality, Modularization, and Language Adaptation	Block II: Multilinguality
  28.6.	L8: Transfer for Token-Level Tasks: Word Alignment & Label Projection	Block II: Multilinguality
  5.7.	L9: Neural Machine Translation	Block III: Advanced
  12.7.	L10: Multilingual Sentence Representations	Block III: Advanced
  19.7.	L11: Prompting and LLMs; ChatGPT/GPT-4	Block III: Advanced

  
  

  
  
• Select topic Exams

  Exams
  • Select activity Exam Results (Klausur, 20.7.2023)

    

    Exam Results (Klausur, 20.7.2023) Fitxer