Dataset Viewer
Auto-converted to Parquet Duplicate
answer
stringclasses
5 values
chart_id
stringclasses
2 values
chart_path
stringclasses
2 values
chart_source
stringclasses
1 value
chart_type
stringclasses
1 value
chart_url_path
stringclasses
2 values
environment
stringclasses
1 value
expected_answer
stringclasses
5 values
hf_chart_url
stringclasses
2 values
id
stringclasses
5 values
label_path
stringclasses
2 values
qa_path
stringclasses
5 values
question
stringclasses
5 values
question_id
stringclasses
3 values
task_config_path
stringclasses
5 values
The letter E has the highest frequency, at 0.12702.
chart_0110_bar
web/AntV G2/bar/chart_0110_bar.html
AntV G2
bar
/web/AntV G2/bar/chart_0110_bar.html
dynamic_chart
The letter E has the highest frequency, at 0.12702.
https://huggingface.co/datasets/chartact-benchmark/ChartAct/resolve/main/web/AntV%20G2/bar/chart_0110_bar.html
chart_0110_bar_q1
label/AntV G2/bar/chart_0110_bar.json
question_answer/AntV G2/bar/PreQA/chart_0110_bar_q1.json
In this dataset, which letter has the highest frequency? What is the corresponding value?
q1
evaluation_examples/examples/chart_data/chart_0110_bar_q1.json
The letters E and T, at 0.12702 and 0.09056 respectively.
chart_0110_bar
web/AntV G2/bar/chart_0110_bar.html
AntV G2
bar
/web/AntV G2/bar/chart_0110_bar.html
dynamic_chart
The letters E and T, at 0.12702 and 0.09056 respectively.
https://huggingface.co/datasets/chartact-benchmark/ChartAct/resolve/main/web/AntV%20G2/bar/chart_0110_bar.html
chart_0110_bar_q2
label/AntV G2/bar/chart_0110_bar.json
question_answer/AntV G2/bar/PreQA/chart_0110_bar_q2.json
Which letters have frequency greater than 0.09? What are their specific values?
q2
evaluation_examples/examples/chart_data/chart_0110_bar_q2.json
The letter T is higher; A is 0.08167, and T is 0.09056.
chart_0110_bar
web/AntV G2/bar/chart_0110_bar.html
AntV G2
bar
/web/AntV G2/bar/chart_0110_bar.html
dynamic_chart
The letter T is higher; A is 0.08167, and T is 0.09056.
https://huggingface.co/datasets/chartact-benchmark/ChartAct/resolve/main/web/AntV%20G2/bar/chart_0110_bar.html
chart_0110_bar_q3
label/AntV G2/bar/chart_0110_bar.json
question_answer/AntV G2/bar/PreQA/chart_0110_bar_q3.json
Between the letters A and T, which one has a higher frequency? What are their respective values?
q3
evaluation_examples/examples/chart_data/chart_0110_bar_q3.json
1957 has the highest sales, at 145.
chart_0111_bar
web/AntV G2/bar/chart_0111_bar.html
AntV G2
bar
/web/AntV G2/bar/chart_0111_bar.html
dynamic_chart
1957 has the highest sales, at 145.
https://huggingface.co/datasets/chartact-benchmark/ChartAct/resolve/main/web/AntV%20G2/bar/chart_0111_bar.html
chart_0111_bar_q1
label/AntV G2/bar/chart_0111_bar.json
question_answer/AntV G2/bar/PreQA/chart_0111_bar_q1.json
In this simple bar chart, which year has the highest sales? What is the sales value?
q1
evaluation_examples/examples/chart_data/chart_0111_bar_q1.json
There are 4 years with sales of 38: 1951, 1959, 1960, and 1962.
chart_0111_bar
web/AntV G2/bar/chart_0111_bar.html
AntV G2
bar
/web/AntV G2/bar/chart_0111_bar.html
dynamic_chart
There are 4 years with sales of 38: 1951, 1959, 1960, and 1962.
https://huggingface.co/datasets/chartact-benchmark/ChartAct/resolve/main/web/AntV%20G2/bar/chart_0111_bar.html
chart_0111_bar_q2
label/AntV G2/bar/chart_0111_bar.json
question_answer/AntV G2/bar/PreQA/chart_0111_bar_q2.json
How many years have sales of 38? Which years are they?
q2
evaluation_examples/examples/chart_data/chart_0111_bar_q2.json

ChartAct Preview

This repository is a preview release of ChartAct, an interactive benchmark for dynamic chart understanding. The full benchmark will be released after formal publication.

Paper: ChartAct: A Benchmark for Dynamic Chart Understanding

GitHub: https://github.com/wulin-wulin/OSWorld_Chart

Contents

  • data/dynamic_chart_preview.jsonl: 5 Dynamic Chart preview samples.
  • data/dashboard_chart_preview.jsonl: 5 Dashboard Chart preview samples.
  • evaluation_examples/: minimal OSWorld-style task configs for the preview samples.
  • web/ and web_insert/: chart webpages required by the preview samples.
  • label/ and question_answer/: metadata and QA annotations for the preview samples.

This preview is intended to show the benchmark format and runnable framework without releasing the full evaluation set before publication.

Loading

from datasets import load_dataset

dataset = load_dataset("chartact-benchmark/ChartAct")
dynamic = dataset["dynamic_chart_preview"]
dashboard = dataset["dashboard_chart_preview"]
print(dynamic[0])

Citation

@misc{chartact2026,
  title={ChartAct: A Benchmark for Dynamic Chart Understanding},
  author={Muye Huang and Lin Wu and Lingling Zhang and Hang Yan and Zhiyuan Wang and Yumeng Fu and Zesheng Yang and Jun Liu},
  year={2026},
  eprint={2605.26994},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}
Downloads last month
18

Paper for chartact-benchmark/ChartAct