FastVLM: Efficient Vision Encoding for Vision Language Models

Benchmark Model Rank Results
visual-question-answering-vqa-on-seed-seedbenchFastVLM-7B#14Accuracy: 75.4
visual-question-answering-on-docvqa-testFastVLM-7B#21ANLS: 0.932
chart-question-answering-on-chartqaFastVLM-7B#451:1 Accuracy: 85.0
optical-character-recognition-ocr-on-ocrbenchFastVLM-7B#59Score: 731
visual-question-answering-vqa-on-realworldqaFastVLM-7B#67Accuracy: 67.2
multimodal-reasoning-on-mmmuFastVLM-7B#109Accuracy: 45.4