List your company

Chart Design Prediction on PewResearch (test)

88AccuracyChartDesign
29.7644.886075.12Apr 6, 2026

Evaluation results

MethodLinks
ChartDesignbackbone=Phi-3, adaptation=Full FT, training_data=Pew+CharXiV2026.0488
ChartDesignbackbone=Qwen-3, adaptation=LoRA, training_data=Pew+CharXiV2026.0484
ChartDesignbackbone=InternVL-2.5, adaptation=LoRA, training_data=Pew+CharXiV2026.0480
ChartDesignbackbone=Phi-3, adaptation=LoRA, training_data=Pew+CharXiV2026.0478
ChartDesignbackbone=InternVL-2.5, adaptation=LoRA, training_data=Pew2026.0475
ChartDesignbackbone=Phi-3, adaptation=LoRA, training_data=Pew2026.0474
ChartDesignbackbone=Qwen-3, adaptation=LoRA, training_data=Pew2026.0472
Phi-3adaptation=Zero-shot, parameters=4B, type=backbone baseline2026.0453
Qwen-3adaptation=Zero-shot, parameters=8B, type=backbone baseline2026.0452
Qwen-3adaptation=Few-shot prompting (3-shot), parameters=8B2026.0451
Llama 3.2adaptation=Zero-shot, parameters=11B, type=base model2026.0444
Llama 3adaptation=Zero-shot, parameters=8B, type=base model2026.0443
Ministral-3adaptation=Zero-shot, parameters=8B, type=base model2026.0442
Mistraladaptation=Zero-shot, parameters=7B, type=base model2026.0441
Gemmaadaptation=Zero-shot, parameters=7B, type=base model2026.0440
Phi-3adaptation=Zero-shot, parameters=4B, type=base model2026.0438
InternLM-3adaptation=Zero-shot, parameters=8B, type=base model2026.0435
DeepSeek-LLMadaptation=Zero-shot, parameters=7B, type=base model2026.0435
InternLM-2.5adaptation=Zero-shot, parameters=8B, type=base model2026.0432
InternLM-2.5adaptation=Zero-shot, parameters=8B, type=backbone baseline2026.0432