List your company
ResearchFormerly WizWand

Multimodal Reasoning on MMT-Bench

57.88AccuracyJigsaw
42.820846.730450.6454.5496May 28, 2024Aug 30, 2024Dec 3, 2024Mar 7, 2025Jun 10, 2025Sep 12, 2025Dec 16, 2025
Updated 8mo ago

Evaluation results

MethodLinks
JigsawPrompting=CoT prompting, Backbone=Qwen-VL-2.5 3B2025.1257.88
MixPrompting=CoT prompting, Backbone=Qwen-VL-2.5 3B2025.1257.72
Jigsaw-R1Prompting=CoT prompting, Backbone=Qwen-VL-2.5 3B2025.1257.53
Qwen 2.5 VLPrompting=CoT prompting, Backbone=Qwen-VL-2.5 3B2025.1253.31
LLaVA-NeXT+CALLLM=Vicuna-13B, Resolution setting=High resolution2024.0552.4
VLM-R1Prompting=CoT prompting, Backbone=Qwen-VL-2.5 3B2025.1252.06
MGM+CALLLM=Vicuna-13B, Resolution setting=Low resolution2024.0551.9
MGM+CALLLM=Vicuna-7B, Resolution setting=Low resolution2024.0551.4
MGM-HD+CALLLM=Vicuna-13B, Resolution setting=High resolution2024.0551.4
LLaVA-1.5+CALLLM=Vicuna-13B, Resolution setting=Low resolution2024.0551.2
LLaVA-1.5LLM=Vicuna-13B, Resolution setting=Low resolution2024.0551.1
MGM-HDLLM=Vicuna-13B, Resolution setting=High resolution2024.0550.9
LLaVA-NeXT+CALLLM=Vicuna-7B, Resolution setting=High resolution2024.0550.7
MGM-HD+CALLLM=Vicuna-7B, Resolution setting=High resolution2024.0550.5
LLaVA-NeXTLLM=Vicuna-13B, Resolution setting=High resolution2024.0550.4
MGMLLM=Vicuna-7B, Resolution setting=Low resolution2024.0550.3
LLaVA-NeXTLLM=Vicuna-7B, Resolution setting=High resolution2024.0549.5
MGMLLM=Vicuna-13B, Resolution setting=Low resolution2024.0549.1
LLaVA-1.5+CALLLM=Vicuna-7B, Resolution setting=Low resolution2024.0548.8
LLaVA-1.5LLM=Vicuna-7B, Resolution setting=Low resolution2024.0548.6
MGM-HDLLM=Vicuna-7B, Resolution setting=High resolution2024.0547.9
MGM+CALLLM=Gemma-2B, Resolution setting=Low resolution2024.0545.4
MGMLLM=Gemma-2B, Resolution setting=Low resolution2024.0543.4