Compute for Robotic Manipulation

Overview

This repository contains the code and data to reproduce the analyses and figures for the Epoch AI Data Insight: Compute is not a bottleneck for robotic manipulation.

The workflow is centered on a single Jupyter notebook, analysis.ipynb, which reads input data from data/.

Repository structure

• analysis.ipynb: End-to-end analysis notebook that reproduces results and figures.
• requirements.txt: Python dependencies to run the notebook.
• data/
  • Robotics Models.csv: Core dataset of robotics models used in the analysis.
  • ML Models-full view.csv: Epoch's AI Models dataset, filtered to frontier AI models in the analysis.
  • robotics_models_docs.md: Definitions of key fields in Robotics Models.csv.

Quick start

• Setup

  # from the repository root (this directory)
  python3 -m venv .venv
  source .venv/bin/activate    # Windows: .venv\Scripts\activate
  pip install -U pip
  pip install -r requirements.txt

• Run the analysis interactively

  jupyter notebook   # or: jupyter lab

  Open analysis.ipynb and Run All cells. Outputs (tables/figures) will appear inline in the notebook. Some cells may optionally save files; if so, they save to the repository directory.

• Run the analysis non-interactively (headless)

  jupyter nbconvert --to notebook --execute --inplace analysis.ipynb

  To generate a shareable HTML report:

  jupyter nbconvert --to html --execute --output analysis.html analysis.ipynb

Citation

If you use this code or data, please follow the instructions to cite the article at https://epoch.ai/data-insights/compute-for-robotic-manipulation.