---
title: "Pick a shape for the job"
lesson_id: "05"
---

# Pick a shape for the job

- **Lesson ID:** 05
- **Goal:** Choose a model family from an app input and output, then follow the path to profiling.
- **Human lesson:** [05-picking-a-shape.html](05-picking-a-shape.html)

## Prerequisites

- Lessons 01 to 04, or a clear distinction between language, classifier, and generative image jobs.
- An app feature that needs a model.

## Inputs, outputs, and artifacts

- **Inputs:** An app feature such as photo labels, conversation, summary, drawing, or image editing.
- **Outputs:** A practical first-choice model shape and a path back to the two published courses.
- **Artifacts:** Progressive chooser `.flow` diagrams and links to the forward pipeline plus From the metal to the model lessons 01 and 07.

## Agent build steps

1. Start with the required output, not a model name.
2. Map photo labels to an image classifier and connect it to [the forward pipeline lesson](../pytorch-core-ai-xcode/01-the-pipeline.html).
3. Map conversation or summary to a language model.
4. Map drawing or pixel editing to a generative image model.
5. Explain that the choice is a first guess that still needs measurement.
6. Link [From the metal to the model lesson 01](../core-ai-from-metal-back/01-metal-shaders.html) and [lesson 07](../core-ai-from-metal-back/07-back-to-the-pipeline.html), plus both course homes.

## Constraints

Keep the chooser practical and plain. No math deep-dives, papers, CUDA, quizzes, or em dashes. Keep both prior-course links relative. Use HTML `.flow` diagrams, not Mermaid.

## Key concepts

- Start from the app input and output, not a model name.
- Photo labels map to an image classifier path, like ClothingCNN.
- Conversation, rewrite, or summary maps to a language model path.
- A new or edited picture needs a generative image path. A classifier cannot draw.

## Takeaways

- You can map a feature to classifier, language model, or generative image model.
- You can follow the forward pipeline or From the metal to the model after you pick a shape.
- You know architecture is a first guess and on-device measurement confirms it.

## Acceptance checks

- Photo labels map to an image classifier.
- Conversation or summary maps to a language model.
- Drawing or editing pixels maps to a generative image model.
- The forward pipeline lesson and From the metal to the model lessons 01 and 07 are linked.
- Both course homes are linked and the Agent brief link is present.
