Fine-Tuning Large Language Models (LLMs) скачать в хорошем качестве

Fine-Tuning Large Language Models (LLMs)

Join us for the first session on fine-tuning large language models (LLMs) tailored for Lightricks developers. We will cover: 1) Fine-Tuning Overview: Understanding of what is fine-tuning, and how it differs from prompt engineering and RAG methods. 2) Live Code Example: Practical demonstration of the fine-tuning process using LLaMA 2-7b-chat LLM. What will you learn? Fine-tuning overview: What is Fine-tuning? When to use Prompt Engineering vs. RAG vs. Fine-tuning? How to Fine-tune an LLM? Fine-tuning code example: How to implement Supervised Fine-tuning (SFT) in code? How to evaluate an LLM? Chapters: 00:00 - About myself 00:42 - Agenda 01:25 - What You'll learn today 02:18 - Part I -- Fine-tuning overview lecture 02:23 - Pre-training 03:44 - What is Fine-tuning? 05:43 - Fine-tuning Example 1 08:15 - Fine-tuning Example 2 09:49 - Why to fine-tune? 10:41 - Prompt Engineering vs. Fine-tuning 14:05 - Benefits of Fine-tuning your own LLM 17:21 - RAG vs. Fine-tuning 20:13 - 3 Ways to Fine-tune 24:20 - RLHF 27:39 - 3 Ways to Parameter Training 28:58 - LLMs are based on the Transformers architecture 33:43 - LoRA 36:38 - Fine-tuning Iterations Process 38:00 - Part II -- Fine-tuning code example 39:34 - LLaMA-2 40:50 - LLaMA-2 vs. GPT-3 42:10 - Preliminaries 44:25 - Quantization 47:46 - Tokenization 49:40 - Inference 52:10 - Data Preparation 58:27 - Training - Supervised Fine Tuning (SFT) 01:02:20 - Saving, Loading and Exporting the model 01:03:41 - Evaluation 01:14:18 - Questions?