snupilab

THETA-Bench Evaluation Artifacts

Private collection of evaluation artifacts for multiple model checkpoints trained on the THETA simulation dataset, organized by agent, run, and snapshot identifiers.

Downloads89
Episodes3003

Why This Matters for Physical AI

Provides systematic evaluation artifacts and benchmarking infrastructure for comparing multiple policy learning models (ACT, DP, GR00T, Psi0, pi0.5, Cosmos, DreamZero, FastWAM) across diverse execution conditions, supporting rigorous assessment of embodied AI algorithms.

Technical Profile

Modalities
videotrajectorymetrics
Environment
simulation
Episodes
3003
Data Format
JSON
Annotation Types
action_labelsreward_labels
Part of the THETA-Bench family

Access

Need custom video data?

Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets