HUGGING FACE DATASET

colosseum-challenge

Name: colosseum-challenge
Creator: colosseum

colosseum/colosseum-challenge — source facts cached from the per-dataset Hugging Face API.

DIRECT ANSWER

License: MIT; size: Not provided by Hugging Face; creator: colosseum; paper: arXiv 2402.08191.

Commercial use: Review the linked license and upstream rights.

API snapshot · Jul 14, 2026

Downloads

355

Hugging Face metadata

Size category

Not provided by Hugging Face

Hugging Face metadata

Last modified

Mar 3, 2025

Hugging Face metadata

Access

Not marked as gated

Per-dataset API

Dataset-card excerpt

Colosseum Dataset Card This dataset contains demonstrations for training and testing Imitation Learning based policies, taken from our simulation benchmark Colosseum, which is based on RLBench. The benchmark consists of 20 tasks from the RLBench suite. We implement variations for each task, like camera pose, which try to test generalization capabilities. Dataset details The training set consits of 100 demonstrations of the 20 tasks without any variation factor…

Hugging Face metadata

Captured tags

Formats: Not provided in the captured tags
Modalities: Not provided in the captured tags

Hugging Face

Open the source dataset

Review the current card, files, access terms, and metadata.

One methodology for every dataset

How this profile is built

See how license, size, creator, downloads, paper links, missing fields, and fetch failures are handled.

Robotics datasets