Qwen-VL

Reported on 4 benchmarks across 3 tasks · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing2 results

Visual Question Answering (VQA)onDocVQA test
ANLS· uses extra data· 2023-08-24
0.651
best: 0.9436 (Human)
Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond arXiv:2308.12966
Visual Question Answering (VQA)onChartQA
1:1 Accuracy· uses extra data· 2023-08-24
65.7
best: 81.3 (ChartPaLI-5B + PaLM 2-S)
Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond arXiv:2308.12966

Reasoning1 result

Natural Language Visual GroundingonScreenSpot
Accuracy (%)· 2023-08-24
5.2
best: 86.34 (UGround-V1-7B)
SOTA
Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond arXiv:2308.12966

Computer Code1 result

Chart Question AnsweringonChartQA
1:1 Accuracy· uses extra data· 2023-08-24
65.7
best: 81.3 (ChartPaLI-5B + PaLM 2-S)
Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond arXiv:2308.12966