Start
Install dagzoo, inspect curated recipes, and generate the first reproducible run.
Documentation
Start guides, recipe packs, artifact contracts, and runtime guides for dagzoo.
dagzoo is easiest to approach in this order: start with named recipes, generate
one run, then drop into the contract and deeper controls only when you need
them.
Start Here
- Start: install, list recipes, generate a first run, and use the PyTorch bridge.
- Reference Packs: curated recipe catalog, confidence tiers, and citation guidance.
- Publish to Hugging Face Hub: generate a handoff root, authenticate, and publish the public corpus.
Go Deeper
- Advanced Controls: custom YAML workflows and deeper command controls.
- How It Works: runtime model and terminology.
- Dagzoo Transform Reference: formal transform math, notation, and operator definitions.
- Feature Guides: diagnostics, interventions, missingness, many-class, shift, steering, noise, and benchmark guardrails.
Contracts
- Artifacts & API: shard layout, metadata schema, handoff manifest, and in-memory sample contract.
Reference Packs
Named dagzoo recipe packs, confidence tiers, and citation guidance.
Publish to Hugging Face Hub
Generate a handoff root, authenticate with Hugging Face, and publish a public or private dataset repo.
How It Works
End-to-end runtime behavior, core concepts, and pipeline walkthrough.
Dagzoo Transform Reference
Formal transform equations, notation, and operator definitions.
Advanced Controls
Advanced controls and custom YAML workflows beyond the curated recipe catalog.
Feature Guides
Guides for diagnostics, missingness, many-class, shift, steering, noise, and benchmark guardrails.
Artifacts & API
Artifact schema, API contract, and handoff metadata layout.
Export Contract Fields
Exhaustive field-by-field export contract catalog generated from the checked-in inventory.