Hugging Face's logo

Datasets:
perpetual-research
/
shadowbench

RL Environment
Tasks:
Reinforcement Learning
Image-Text-to-Text
Modalities:
Image
Tabular
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
Tags:
benchmark
spatial-reasoning
embodied-reasoning
hands
shadows
agents
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
shadowbench / data
382 MB
  • 1 contributor
 History: 3 commits
spratyey's picture
spratyey
ShadowBench: every call of all 360 episodes, with the images the models received
7ada62c verified