The same prompt can give very different results depending on the model and the settings behind it. Model parameters are the dials you send with each request that shape how the model picks its words, while model selection is choosing which model should do the job in the first place.
You'll start with the sampling settings: temperature, which controls how random the output is, and top-p and top-k, which limit which words the model can choose from. Then you'll learn how to compare models on cost, speed, and quality so you can match the right model to each task.
Once you can control and choose your model, you'll want to give it the right information to work with. That's the focus of RAG & Context Engineering, the next module in this course.
A practical look at how temperature, top-p, and top-k change what a model writes, and how to tune them.
How the main LLM sampling strategies work and how each setting shapes the model's next-word choices.
The key generation settings, including min-p and repeat penalties, and what each one does to the output.
How to compare different language models on the same prompt so you can pick the one that fits your task.