datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ChartEdit
ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs' Capability via Chart Editing
Paper | Code
ChartEdit is a novel benchmark designed for chart editing tasks, proposed in the paper "ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs' Capability via Chart Editing". It features $1405$ diverse editing instructions applied to $233$ real-world charts, each manually annotated and validated for accuracy. This benchmark aims to evaluate… See the full description on the dataset page: https://huggingface.co/datasets/xxxllz/ChartEdit.chart
Format
There is no single top-level annotations.jsonl in this directory. Instead, the annotations are split across these 5 subdirectories:
clean_TSQA/anomaly_detection/annotations.jsonl
clean_TSQA/classification/annotations.jsonl
clean_TSQA/open_ended_qa/annotations.jsonl
clean_cats-bench_hard/annotations.jsonl
clean_timeomni/annotations.jsonl
Each line is one JSON object (JSONL format). The core fields are described below.
Common Fields
id (string)… See the full description on the dataset page: https://huggingface.co/datasets/tlhuang/chart.