PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models

Benchmark Model Rank Results
visual-question-answering-vqa-on-seed-seedbenchPerceptionDLM-Base (8B)#4Accuracy: 78.9
chart-question-answering-on-chartqaPerceptionDLM-Base#51:1 Accuracy: 91.6
visual-question-answering-vqa-on-mmbench-enPerceptionDLM-Base (8B)#16Accuracy: 85.0
visual-question-answering-on-v-benchPerceptionDLM-Base (8B)#19Accuracy: 73.3
visual-question-answering-vqa-on-blinkPerceptionDLM-Base (8B)#26Accuracy: 60.3
visual-question-answering-vqa-on-realworldqaPerceptionDLM-Base (8B)#30Accuracy: 73.7
vision-language-understanding-on-mmstarPerceptionDLM-Base (8B)#69Accuracy: 63.7
multimodal-reasoning-on-mathvistaPerceptionDLM-Base (8B)#80Accuracy: 65.5
multimodal-reasoning-on-mmmuPerceptionDLM-Base (8B)#102Accuracy: 47.2