YSDA Natural Language Processing Course

repository·2025·Indexed 27 days ago

https://github.com/yandexdataschool/nlp_course

Materials for the 2025 iteration of the YSDA Natural Language Processing course. The 14-week curriculum covers modern NLP and LLM technologies, including word embeddings, Transformers, Large Language Models, PEFT, RLHF, RAG, AI agents, and multimodal LLMs. The repository includes lectures, seminar notebooks, and homework assignments.

Tokens
33.3K
Snippets
72
Records
169
Agent score
94%

What's inside yandexdataschool-nlp_course

  1. Overview of Building LLM Systems (Week 13)

    2025

    This module covers the transition of Large Language Model (LLM) systems from initial prototypes to production-ready environments. Key topics include:

    • Designing production pipelines from prototypes.
    • System architecture considerations for LLM applications.
    • Strategies for scaling and optimizing LLM performance.
    • Addressing real-world deployment challenges.
  2. Overview of AI Agents in Production (Week 14)

    2025

    This module covers the deployment of AI agents in production environments. Key topics include:

    • Production-ready agent architectures
    • Monitoring and observability
    • Handling edge cases and failures
    • Industry best practices

    Note: This is an advanced topic. For foundational concepts, refer to Week 10: AI Agents and the seminar on Long-term Agent Memory.

  3. Overview of YSDA Natural Language Processing course

    2025
    The YSDA Natural Language Processing course (2025 iteration) provides lecture and seminar materials covering a wide range of NLP topics, from word embeddings to AI agents in production. Materials are organized into weekly folders (e.g., ./week01_embeddings).
  4. Access Week 03 Attention course materials

    2025

    The Week 03 Attention module provides lecture slides, video recordings, and supplementary research materials.

    Core Materials:

    • Lecture Slides: Available via Google Drive.
    • Course Videos (Russian): Includes both the lecture and the seminar recordings.
    • External Lectures (English): Stanford lectures on seq2seq and MT, and CMU lectures on seq2seq and attention.

    Supplementary Reading & Research:

    • RNNs: Distill.pub post on attention and augmented RNNs, Seq2seq video lectures, and research papers on BLEU and CIDEr metrics.
    • Transformers: The 'Illustrated Transformer' guide.
    • Image Captioning: MSCOCO captioning challenge and a baseline notebook.
  5. Access Week 04: Transfer Learning materials

    2025

    The Week 04 module covers Transfer Learning. You can access the following resources:

  6. Access Machine Translation course materials

    2025

    The Machine Translation module covers word alignment models, noisy channel models, phrase-based MT, and Neural Machine Translation (NMT). The following resources are available:

  7. Access lecture slides

    2025
    This folder contains the slides used during on-campus YSDA lectures. Note that these slides are also referenced within the specific ./week*/ directories. For the best learning experience, it is recommended to browse the course content via the ./week* folders rather than navigating the slides folder directly.
  8. Access Week 06: Prompting & In-Context Learning materials

    2025

    The Week 06 module covers Prompting and In-Context Learning. You can access the following resources:

    • Lecture Slides: Available as a PDF (./lecture_prompting.pdf) or via Google Slides.
    • Video Lectures (Russian): Lecture and seminar videos are hosted on Yandex Disk.
    • English Resources: Includes a lecture on reasoning and a Google MOOC on prompting.
    • Extra Reading: Includes the 'Chain of Thought' paper by Wei et al. (2022) and guides on prompt injection/jailbreaking.