Introduces a benchmark for MLLM-based chart data extraction from unlabeled images and a human-centered training framework that reaches SOTA numerical accuracy with a 7B model.
3 Steven F Roth, John Kolojejchick, Joe Mattis, and Jade Goldstein
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.HC 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
ChartDesign post-trains LLMs on data-design pairs from PewResearch and CharXiV charts to output design attributes like chart type and layout from tabular input, reaching 84% accuracy versus 53% for baselines and producing human-preferred renders.
citing papers explorer
-
Making Multimodal LLMs Reliable Chart Data Extractors: A Benchmark and Training Framework
Introduces a benchmark for MLLM-based chart data extraction from unlabeled images and a human-centered training framework that reaches SOTA numerical accuracy with a 7B model.
-
ChartDesign: Towards LLM Designer of Data Visualization
ChartDesign post-trains LLMs on data-design pairs from PewResearch and CharXiV charts to output design attributes like chart type and layout from tabular input, reaching 84% accuracy versus 53% for baselines and producing human-preferred renders.