Section: Ressources | Modélisation des langues

CM 01 - Static Semantic Representation File

Lab: play with short, static and dense word representations URL

CM 02 - Transformer architecture File

Lab - comprendre l'architecture du Transformer Page

CM 03 - Deeper with Transformers File

Lab: Model size URL

Lab: Generative sampling techniques URL

CM 04 - Pre-training, Prompting or Fine-tuning LLMs File

Lab: Instructed model and prompting (zero and few shots) URL

CM 05 - Reducing learning and inference costs File

Lab: full vs parameter efficient fine-tuning URL

CM 06 - Aligning language models with human preferences File

Lab: Direct Preference Optimization (DPO) Using Hugging Face URL

Lab: Reinforcement Learning from Human Feedback Using PPO URL

Section outline