List your company

Chart Design Prediction on Pew+CharXiV (test)

82AccuracyChartDesign
21.6837.345368.66Apr 6, 2026

Evaluation results

MethodLinks
ChartDesignbackbone=Phi-3, adaptation=Full FT, training_data=Pew+CharXiV2026.0482
ChartDesignbackbone=Qwen-3, adaptation=LoRA, training_data=Pew+CharXiV2026.0480
ChartDesignbackbone=Phi-3, adaptation=LoRA, training_data=Pew+CharXiV2026.0472
ChartDesignbackbone=InternVL-2.5, adaptation=LoRA, training_data=Pew+CharXiV2026.0470
ChartDesignbackbone=InternVL-2.5, adaptation=LoRA, training_data=Pew2026.0462
ChartDesignbackbone=Qwen-3, adaptation=LoRA, training_data=Pew2026.0461
ChartDesignbackbone=Phi-3, adaptation=LoRA, training_data=Pew2026.0460
Qwen-3adaptation=Few-shot prompting (3-shot), parameters=8B2026.0446
Llama 3.2adaptation=Zero-shot, parameters=11B, type=base model2026.0432
Phi-3adaptation=Zero-shot, parameters=4B, type=backbone baseline2026.0432
Qwen-3adaptation=Zero-shot, parameters=8B, type=backbone baseline2026.0431
Ministral-3adaptation=Zero-shot, parameters=8B, type=base model2026.0430
Llama 3adaptation=Zero-shot, parameters=8B, type=base model2026.0430
Mistraladaptation=Zero-shot, parameters=7B, type=base model2026.0429
InternLM-3adaptation=Zero-shot, parameters=8B, type=base model2026.0428
Gemmaadaptation=Zero-shot, parameters=7B, type=base model2026.0428
Phi-3adaptation=Zero-shot, parameters=4B, type=base model2026.0426
DeepSeek-LLMadaptation=Zero-shot, parameters=7B, type=base model2026.0425
InternLM-2.5adaptation=Zero-shot, parameters=8B, type=base model2026.0424
InternLM-2.5adaptation=Zero-shot, parameters=8B, type=backbone baseline2026.0424