Documentation

This documentation is organized into three sections. Basic: Quickstart guides to get you started. SLURM: Infrastructure for large-scale experiments. In-depth: Details you do not need now, but may want to know later.

The basic section is sufficient to get started with the local data collection, training and evaluation loop.

Basic

• Collect Data
  • Start Data Collection Locally
  • Inspect Collected Data
  • About the Expert
• Training
  • Prerequisites
  • Model Training
  • Common issues
• Evaluation
  • Overview
  • Running Evaluations
  • Summarize Leaderboard 2.0 Results for Longest6 v2 and Town13
  • Summarize Bench2Drive Results
  • Best Practices
• Tutorial Jupyter Notebooks
  • Pipline Verification Notebooks
  • Data Format Explained Interactively
  • Debug Closed-Loop Evaluation Interactively

SLURM

• Overview
  • Why a wrapper?
  • Principle of the wrapper
  • Example
• Collect Data
  • Quick Start
  • Monitor Progress
  • Delete Failed Routes
• Training
  • Overview
  • Pre-training
  • Post-training
  • Resume crashed training
• Evaluation
  • Overview
  • Example

In-Depth

• Inspecting Pickle Files
• Cross-dataset Training
• Configuration System
• Debug Expert

Miscellaneous

• FAQ
• Glossary
• Known Issues