Skip to main content
Ctrl+K
ToyBlocks  documentation - Home ToyBlocks  documentation - Home
  • Workflows
  • GitHub
  • Workflows
  • GitHub

Section Navigation

  • ToyBlocks Installation and Setup Instructions
  • Automatic Data Generation
    • Stage 1: Task generation
    • Stage 2: Environment generation
    • Stage 3: Execution (vecgen)
    • Stage 4: Rendering
    • Stage 5: Merge
    • Stage 6: Annotation
    • Recorded Dataset Fields
    • Running Pipeline Locally
  • Training ManiFlow Policies on Automatically Generated Data
    • train_maniflow.py Reference
    • Setting up a new policy server
  • Policy Evaluation
    • Replaying Actions from a Dataset
    • Eval Metrics
    • launch_eval.py Reference
    • evaluate.py Reference
  • Agentic Robotics
  • Workflows

ToyBlocks

Workflows#

1 · Installation and Setup
  • Prerequisites

  • Setup

  • Download the assets

  • Verify the installation

  • Docker images

2 · Automatic Data Generation
  • Running Pipeline Locally

  • Stages: 1 Tasks · 2 Environments · 3 Execution · 4 Rendering · 5 Merge · 6 Annotation

  • Recorded Dataset Fields

3 · Training ManiFlow Policies
  • Training on a generated dataset

  • Choosing GPUs and nodes

  • Resuming and fine-tuning

  • train_maniflow.py Reference

  • Setting up a new policy server

4 · Policy Evaluation
  • Quick start with launch_eval.py

  • Replaying Actions from a Dataset

  • Eval Metrics

  • launch_eval.py Reference · evaluate.py Reference

Agentic Robotics

Setup · Run an episode · Web UI · Create a new task · ManiFlow skills

previous

ToyBlocks Docs

next

ToyBlocks Installation and Setup Instructions

Show Source

© Copyright 2026, Allen AI.

Created using Sphinx 9.0.4.

Built with the PyData Sphinx Theme 0.18.0.