Module 3 Wrap-up: The Model Comparison Challenge
Put your knowledge to the test. Compare Llama, Mistral, and Gemma on speed, humor, and logic.
Module 3 Wrap-up: The Great Local AI Bake-off
You now know how to navigate the registry, decode tags, and pick the perfect model. Now it’s time for some science. We are going to benchmark the "Big Three" of the local AI world on your specific hardware.
Hands-on Exercise: The Triple Crown Test
We are going to compare Llama 3, Mistral, and Gemma 2 (9B).
1. Download the Competitors
Open your terminal and pull all three:
ollama pull llama3
ollama pull mistral
ollama pull gemma2:9b
2. The Speed Test (Token Generation)
Run each model and ask: "Write a 3-paragraph story about a robot on Mars."
- Observation: Which one started typing first? Which one finished the fastest?
3. The Logic Test (Math/Reasoning)
Ask each model this riddle: "If I have 3 apples and I give you two, how many apples do you have?" (Note: Some models fail this simple logic!)
4. The Coding Test
Ask each: "Write a Python function to check if a string is a palindrome."
- Observation: Which one used the most elegant code? Did any of them fail to explain their work?
The Comparison Matrix
Fill this out for your own machine:
| Feature | Llama 3 (8B) | Mistral (7B) | Gemma 2 (9B) |
|---|---|---|---|
| Download Size | ~4.7 GB | ~4.1 GB | ~5.4 GB |
| First Word Speed | |||
| Logic (Apple Riddle) | |||
| Storytelling Quality |
Module 3 Summary
- The Ollama Registry is your source for all models.
- Tags define the intelligence level and hardware requirement of a model.
- Llama, Mistral, and Gemma each have unique "personalities" and strengths.
- Smaller models (8B) are often superior to larger models if you need extreme speed.
Coming Up Next...
In Module 4, we dive deep into the math. We will learn how Quantization works, what a Transformer actually is, and why the GGUF format is the standard for local AI.
Module 3 Checklist
- I have pulled at least 3 different models.
- I understand the difference between
8band70b. - I know how to check the specific tag of a model.
- I have performed my first "hands-on" comparison.
- I know which model I prefer for creative writing.