У нас вы можете посмотреть бесплатно What is a Transformer? (Transformer Walkthrough Part 1/2) или скачать в максимальном доступном качестве, видео которое было загружено на ютуб. Для загрузки выберите вариант из формы ниже:
Если кнопки скачивания не
загрузились
НАЖМИТЕ ЗДЕСЬ или обновите страницу
Если возникают проблемы со скачиванием видео, пожалуйста напишите в поддержку по адресу внизу
страницы.
Спасибо за использование сервиса ClipSaver.ru
See part 2 here: Implementing GPT-2 from Scratch https://neelnanda.io/transformer-tuto... Template notebook: https://neelnanda.io/transformer-temp... Solution notebook: https://neelnanda.io/transformer-solu... If you enjoyed this, I expect you'd enjoy learning more about what's actually going on inside these models and how to reverse engineer them! Check out: A Comprehensive Mechanistic Interpretability Explainer & Glossary: https://www.neelnanda.io/glossary Concrete Steps for Getting Started in Mechanistic Interpretability: https://www.neelnanda.io/getting-started 200 Concrete Open Problems in Mechanistic Interpretability: https://www.neelnanda.io/concrete-ope... Further resources: The transformers section of my MI explainer: https://dynalist.io/d/n2ZWtnoYHrU1s4v... My TransformerLens library for doing mechanistic interpretability research on GPT-2 style language models: https://github.com/neelnanda-io/Trans... My walkthrough of A Mathematical Framework for Transformer Circuits, for a deeper dive into how to think about transformers: • A Walkthrough of A Mathematical Framework ... Check out these other intros to transformers for another perspective: Jay Alammar's illustrated transformer: https://jalammar.github.io/illustrate... Andrej Karpathy's MinGPT: https://github.com/karpathy/minGPT Timestamps 00:00 Intro 03:12 Setup 07:38 What is the point of a transformer? 12:48 Tokens - Transformer Inputs 25:42 Logits – Transformer Outputs 36:29 Implementation/High level architecture/Embedding 39:08 Attention 45:03 MLPs 48:54 Unembedding 50:43 LayerNorm 59:16 Positional Information