From the uDemy course on LLM engineering.
https://www.udemy.com/course/llm-engineering-master-ai-and-large-language-models
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
245 lines
8.8 KiB
245 lines
8.8 KiB
{ |
|
"cells": [ |
|
{ |
|
"cell_type": "markdown", |
|
"id": "fe12c203-e6a6-452c-a655-afb8a03a4ff5", |
|
"metadata": {}, |
|
"source": [ |
|
"# End of week 1 exercise\n", |
|
"\n", |
|
"To demonstrate your familiarity with OpenAI API, and also Ollama, build a tool that takes a technical question, \n", |
|
"and responds with an explanation. This is a tool that you will be able to use yourself during the course!" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 22, |
|
"id": "c1070317-3ed9-4659-abe3-828943230e03", |
|
"metadata": {}, |
|
"outputs": [], |
|
"source": [ |
|
"# imports\n", |
|
"import os\n", |
|
"import json\n", |
|
"from dotenv import load_dotenv\n", |
|
"from IPython.display import Markdown, display, update_display\n", |
|
"from openai import OpenAI\n", |
|
"import ollama" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 2, |
|
"id": "4a456906-915a-4bfd-bb9d-57e505c5093f", |
|
"metadata": {}, |
|
"outputs": [], |
|
"source": [ |
|
"# constants\n", |
|
"\n", |
|
"MODEL_GPT = 'gpt-4o-mini'\n", |
|
"MODEL_LLAMA = 'llama3.2'" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 10, |
|
"id": "93e65a84", |
|
"metadata": {}, |
|
"outputs": [ |
|
{ |
|
"name": "stdout", |
|
"output_type": "stream", |
|
"text": [ |
|
"API key looks good so far\n" |
|
] |
|
} |
|
], |
|
"source": [ |
|
"# set up environment\n", |
|
"\n", |
|
"load_dotenv(override=True)\n", |
|
"api_key = os.getenv('OPENAI_API_KEY')\n", |
|
"\n", |
|
"if api_key and api_key.startswith('sk-proj-') and len(api_key)>10:\n", |
|
" print(\"API key looks good so far\")\n", |
|
"else:\n", |
|
" print(\"There might be a problem with your API key? Please visit the troubleshooting notebook!\")\n", |
|
" \n", |
|
"openai = OpenAI()" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 19, |
|
"id": "3f0d0137-52b0-47a8-81a8-11a90a010798", |
|
"metadata": {}, |
|
"outputs": [], |
|
"source": [ |
|
"# system prompt\n", |
|
"\n", |
|
"system_prompt = \"\"\"\n", |
|
"You are an assistant that answers questions concisely with technical details. The answer should be as descriptive as possible,\n", |
|
"using language that can be understood by a 5 year old.\n", |
|
"\"\"\"\n", |
|
"\n", |
|
"# here is the question; type over this to ask something new\n", |
|
"\n", |
|
"user_prompt = \"\"\"\n", |
|
"Please explain what this code does and why:\n", |
|
"yield from {book.get(\"author\") for book in books if book.get(\"author\")}\n", |
|
"\"\"\"" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 18, |
|
"id": "60ce7000-a4a5-4cce-a261-e75ef45063b4", |
|
"metadata": {}, |
|
"outputs": [ |
|
{ |
|
"data": { |
|
"text/markdown": [ |
|
"Alright! Imagine you have a magical treasure chest (that’s your code!) filled with sparkly books, and each book has a tag telling you who wrote it—the author! Now, let’s break down what’s happening:\n", |
|
"\n", |
|
"1. **Books Galore**: Think of `books` as a big shelf where all your storybooks are sitting in a neat line. Some books say who the author is, and some books are a bit shy and just sit there without telling you.\n", |
|
"\n", |
|
"2. **Searching for Authors**: The part `book.get(\"author\")` is like asking each book, \"Hey, who wrote you?\" If the book says \"Me! I'm written by Captain Awesome!\" then yay! You found an author! If the book goes “Umm... I forgot!”, you just move on to the next one.\n", |
|
"\n", |
|
"3. **Creating a List**: The fancy `{ ... for book in books if book.get(\"author\")}` part means you're gathering all the names of authors from the books that told you who they are. It's like putting a name sticker on all the books that have an author! \n", |
|
"\n", |
|
"4. **Sparkling Unique Names**: The curly braces `{}` mean you’re creating something special—a *set*! This is like a group of friends where nobody can have the same name, so if two books are written by Captain Awesome, he only gets one sticker!\n", |
|
"\n", |
|
"5. **The Magic of Yielding**: Finally, the `yield from` part is like saying, \"Okay, magic treasure chest, please start handing me those names one by one!\" It allows you to take each author’s name out, one at a time, as if you’re pulling out candy from a piñata—sweet and delightful!\n", |
|
"\n", |
|
"So, to sum it up with a giggle: This code is like a magical helper that goes through your book shelf, gathers all the brave authors who shouted their names, puts them in a special list without repeats, and then hands them out to you one by one like a candy parade! 🍭✨" |
|
], |
|
"text/plain": [ |
|
"<IPython.core.display.Markdown object>" |
|
] |
|
}, |
|
"metadata": {}, |
|
"output_type": "display_data" |
|
} |
|
], |
|
"source": [ |
|
"# Get gpt-4o-mini to answer, with streaming\n", |
|
"\n", |
|
"def get_answer(question):\n", |
|
" stream = openai.chat.completions.create(\n", |
|
" model=MODEL_GPT,\n", |
|
" messages=[\n", |
|
" {\"role\": \"system\", \"content\": system_prompt},\n", |
|
" {\"role\": \"user\", \"content\": question}\n", |
|
" ],\n", |
|
" stream=True\n", |
|
" )\n", |
|
" \n", |
|
" response = \"\"\n", |
|
" display_handle = display(Markdown(\"\"), display_id=True)\n", |
|
" for chunk in stream:\n", |
|
" response += chunk.choices[0].delta.content or ''\n", |
|
" response = response.replace(\"```\",\"\").replace(\"markdown\", \"\")\n", |
|
" update_display(Markdown(response), display_id=display_handle.display_id)\n", |
|
"\n", |
|
"get_answer(user_prompt)" |
|
] |
|
}, |
|
{ |
|
"cell_type": "code", |
|
"execution_count": 24, |
|
"id": "8f7c8ea8-4082-4ad0-8751-3301adcf6538", |
|
"metadata": {}, |
|
"outputs": [ |
|
{ |
|
"name": "stdout", |
|
"output_type": "stream", |
|
"text": [ |
|
"This line of code is doing something really cool with a bunch of books!\n", |
|
"\n", |
|
"Let's break it down:\n", |
|
"\n", |
|
"- `books` is like a big box full of books.\n", |
|
"- Each book is like a small container that has some information inside, like the author's name.\n", |
|
"\n", |
|
"Here's what this line does:\n", |
|
"\n", |
|
"1. It takes all the books in the box (`for book in books`).\n", |
|
"2. For each book, it checks if the book has an \"author\" inside (like a small treasure chest with a key: `\"author\"`).\n", |
|
"3. If a book does have an author (like finding a treasure), it gets the author's name from that book.\n", |
|
"4. It then makes the author's name available to someone else who is waiting for more information, like a helper.\n", |
|
"\n", |
|
"The magic part is `yield from`. Imagine you're on a treasure hunt with your friend. You find a clue and say \"Look! This is a map!\" Then, your friend can look at the map and see where the next clue is. That's kind of what `yield from` does: it says \"Hey, I found this information, here's more info about where to go next\"!\n", |
|
"\n", |
|
"In code terms, when you run this line, it will make each author's name available one by one, like a treasure chest with clues. \n", |
|
"\n", |
|
"Here is an example using python classes:\n", |
|
"\n", |
|
"```python\n", |
|
"class Book:\n", |
|
" def __init__(self, title, author):\n", |
|
" self.title = title\n", |
|
" self.author = author\n", |
|
"\n", |
|
"books = [\n", |
|
" Book(\"Harry Potter\", \"J.K. Rowling\"),\n", |
|
" Book(\"The Lion King\", \"Alexander Dumas\"),\n", |
|
"]\n", |
|
"\n", |
|
"def find_authors(books):\n", |
|
" for book in books:\n", |
|
" if 'author' in book.__dict__:\n", |
|
" yield from (book.author,)\n", |
|
"\n", |
|
"# Test the function\n", |
|
"for author in find_authors(books):\n", |
|
" print(author)\n", |
|
"```" |
|
] |
|
} |
|
], |
|
"source": [ |
|
"# Get Llama 3.2 to answer\n", |
|
"\n", |
|
"# Using ollama python package\n", |
|
"\n", |
|
"def get_answer(question):\n", |
|
" ollama_response = ollama.chat(\n", |
|
" model=MODEL_LLAMA,\n", |
|
" messages=[\n", |
|
" {\"role\": \"system\", \"content\": system_prompt},\n", |
|
" {\"role\": \"user\", \"content\": question}\n", |
|
" ],\n", |
|
" stream=True\n", |
|
" )\n", |
|
" \n", |
|
" # Printing out each piece of the generated response while preserving order\n", |
|
" for chunk in ollama_response:\n", |
|
" print(chunk['message']['content'], end='', flush=True)\n", |
|
"\n", |
|
"\n", |
|
"get_answer(user_prompt)\n" |
|
] |
|
} |
|
], |
|
"metadata": { |
|
"kernelspec": { |
|
"display_name": "venv", |
|
"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": "3.11.11" |
|
} |
|
}, |
|
"nbformat": 4, |
|
"nbformat_minor": 5 |
|
}
|
|
|