Tutorials & Cookbooks

This section collects walkthroughs and recipes. It will expand as new examples and cookbooks are added.

Examples

These examples provide hands-on introductions to core plait concepts:

Example	Description
01_module.py	Module basics: creating custom modules, composing hierarchies, using Parameters
02_llm_pipeline.py	LLM pipelines: sequential, parallel fan-out, fan-in synthesis patterns
03_tracing.py	Tracing: capturing execution DAGs, inspecting nodes and dependencies
04_execution.py	Execution: run(), bind(), ExecutionSettings, batch processing
05_optimization.py	Optimization: train/eval modes, loss functions, backward passes

Cookbooks

Cookbooks demonstrate complete end-to-end workflows:

Cookbook	Description
hallucination_detection.py	Training a hallucination detector with HaluBench dataset

Suggested Learning Path

1. Start with 01_module.py to understand the basic building blocks
2. Learn LLM pipeline patterns in 02_llm_pipeline.py
3. Understand how plait captures computation graphs with 03_tracing.py
4. See how to execute pipelines in 04_execution.py
5. Learn optimization techniques in 05_optimization.py
6. Apply your knowledge with the hallucination detection cookbook