Dataset · opendatalab
Chart and diagram reasoning SFT dataset
opendatalab/ChartVerse-SFT-1.8M
1.8M chart QA pairs with Chain-of-Thought annotations for VLM training.
What's inside
ChartVerse-SFT-1800K is a large dataset for chart and diagram reasoning with Chain-of-Thought annotations, part of the opendatalab/ChartVerse project. It contains all verified correct examples without error rate filtering: unlike SFT-600K, it includes easy examples (r=0), providing a full set of ground-truth QA pairs. Modalities are image and text, language is English, size is 1M to 10M.
What you can do with it
- Train VLMs (SFT) to reason over charts with CoT
- Fine-tune LoRA for chart reasoning
- Evaluate models on visual chart QA
- Tasks
- Visual question answering, Image text → text
- Languages
- English
- Size
- 1M–10M rows
- Downloads this month
- 4.7K
- Likes
- 141
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("opendatalab/ChartVerse-SFT-1.8M")