ZAYA1-VL-8B Technical Report

Benchmark Model Rank Results
visual-question-answering-vqa-on-infographicvqaZAYA1-VL-8B#11ANLS: 74.0
visual-grounding-on-refcoco-refcoco-refcocog-overallZAYA1-VL-8B#18Accuracy: 84.3
visual-question-answering-vqa-on-seed-seedbenchZAYA1-VL-8B#18Accuracy: 72.7
visual-question-answering-vqa-on-textvqa-valZAYA1-VL-8B#21Accuracy: 74.4
visual-question-answering-on-docvqa-testZAYA1-VL-8B#29ANLS: 0.925
visual-question-answering-vqa-on-blinkZAYA1-VL-8B#46Accuracy: 45.9
optical-character-recognition-ocr-on-ocrbenchZAYA1-VL-8B#53Score: 798
chart-question-answering-on-chartqaZAYA1-VL-8B#581:1 Accuracy: 82.2
visual-question-answering-vqa-on-realworldqaZAYA1-VL-8B#72Accuracy: 65.0
multimodal-reasoning-on-mathvistaZAYA1-VL-8B#84Accuracy: 64.0
multimodal-reasoning-on-mmmuZAYA1-VL-8B#107Accuracy: 46.0