-
Notifications
You must be signed in to change notification settings - Fork 4
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
6 changed files
with
294 additions
and
111 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,127 @@ | ||
| { | ||
| "cells": [ | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "# 400.9 | Fine Tuning Large Language Models\n", | ||
| "\n", | ||
| " **This notebook is for my personal use only** - all sources are cited. If you are following the same learning journey please reference original sources instead. Key Resources used include:\n", | ||
| "1. [Fine Tuning Large Language Models](https://www.deeplearning.ai/short-courses/finetuning-large-language-models/), _DeepLearning.AI_ (2024)" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "---\n", | ||
| "\n", | ||
| "## Learn: Concepts\n" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "### What is Fine Tuning?\n", | ||
| "\n", | ||
| "Process of turning a general-purpose pre-trained model into a speciliazed version suited for a particular task. Analogy: a general practitioner (GP) vs. a specialist (cardiologist).\n", | ||
| "\n", | ||
| "Both fine-tuning and prompt-engineering are techniques to improve the quality of a model's response to a user request but differ in cost and context:\n", | ||
| " - Prompt engineering is easier to implement, has less upfront cost\n", | ||
| " - Prompt engineering has data limitations (fewer examples), more hallucinations\n", | ||
| " - Fine-tuning is more effective but has upfront compute & data processing costs\n", | ||
| " - Fine-tuning requires high-quality data and more expertise in model training\n", | ||
| "\n", | ||
| "### Why Fine Tune?\n", | ||
| "1. More cost-effective - (per-request) frees up space used by examples, context\n", | ||
| "1. More consistent outputs - understands app requirements, response formats\n", | ||
| "1. Reduce hallucinations - grounded in relevant data, critical for enterprise\n", | ||
| "1. Improve data privacy - reduce breaches, data leakage in training\n", | ||
| "1. Better performance - reliability, lower latency, better moderation options" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "---\n", | ||
| "\n", | ||
| "## Apply: Tasks\n", | ||
| "\n", | ||
| "> The following exercises should help walk through the entire process of fine-tuning a large language model using a specific provider and model endpoint." | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "\n", | ||
| "\n", | ||
| "### T1: Setup Dev Environment\n", | ||
| "\n", | ||
| "To explore ideas in practice, we need access to a relevant Large Language Model (LLM) and provider-hosted endpoint (API). Use the [LLM Setup](./400-00-aoia-intro.ipynb) notebook to configure environment variables and validate setup for supported providers including:\n", | ||
| " - Open AI\n", | ||
| " - Azure Open AI\n", | ||
| " - Hugging Face\n" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "---\n", | ||
| "\n", | ||
| "### T2: Lamini Example\n", | ||
| "\n", | ||
| "The example from the DeepLearning.AI course uses the following libraries:\n", | ||
| "- PyTorch (Meta) - lowest level\n", | ||
| "- Transformers (Hugging Face) - abstracts PyTorch for easier use\n", | ||
| "- Llama (Lamini) - abstracts working with LLama models\n" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "code", | ||
| "execution_count": null, | ||
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "# Install the lamini package\n", | ||
| "import lamini\n", | ||
| "import os\n", | ||
| "lamini.api_key = os.getenv(\"LAMINI_API_KEY\")\n", | ||
| "\n", | ||
| "# Test the installation\n", | ||
| "from llama import BasicModelRunner\n", | ||
| "non_ft_model = BasicModelRunner(\"meta-llama/LLama-3-7b-hf\")\n", | ||
| "print(non_ft_model(\"Oh say can you see\"))" | ||
| ] | ||
| }, | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, | ||
| "source": [] | ||
| } | ||
| ], | ||
| "metadata": { | ||
| "kernelspec": { | ||
| "display_name": "Python 3", | ||
| "language": "python", | ||
| "name": "python3" | ||
| }, | ||
| "language_info": { | ||
| "codemirror_mode": { | ||
| "name": "ipython", | ||
| "version": 3 | ||
| }, | ||
| "file_extension": ".py", | ||
| "mimetype": "text/x-python", | ||
| "name": "python", | ||
| "nbconvert_exporter": "python", | ||
| "pygments_lexer": "ipython3", | ||
| "version": "3.10.13" | ||
| } | ||
| }, | ||
| "nbformat": 4, | ||
| "nbformat_minor": 2 | ||
| } |
File renamed without changes.
Oops, something went wrong.