dmarsili/ExpertLens-Qwen3-VL-30B-A3B
31B • Updated • 11
Datasets:
data_source stringclasses 1
value | prompt listlengths 2 2 | images listlengths 1 1 | ability stringclasses 1
value | reward_model dict | extra_info dict |
|---|---|---|---|---|---|
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAhoAAAIFCAIAAAAbW8M/AAEAAElEQVR4nMT96ZcdN3IngAaQeffaNxY3cRepXe1ut(...TRUNCATED) | math_vqa | {
"ground_truth": "$$35^{\\circ}$$",
"style": "rule"
} | {"answer":"$$35^{\\circ}$$","id":"abb5ab16-df73-44d9-8869-405babf18eeb","index":3073,"question":"As (...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAADSkAAAHaCAIAAADZ57qlAAEAAElEQVR4nOz9SY9tSZIeCIqo6hnubOObfIgpkxlJd(...TRUNCATED) | math_vqa | {
"ground_truth": "$$0.6$$",
"style": "rule"
} | {"answer":"$$0.6$$","id":"ea991007-c539-4357-9306-a353da5212a7","index":1176,"question":"The table b(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAGwAAABZCAYAAADb0cHTAAAPLUlEQVR4nO2de0xUxxfHP3d3YWFRwJVHKvFFWi2og(...TRUNCATED) | math_vqa | {
"ground_truth": "3/4",
"style": "rule"
} | {"answer":"3/4","id":"1b879781-05b3-4254-b1cf-34d64ffb3275","index":1533,"question":"As shown in the(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAK0AAABjCAYAAADpTYCHAAAd8UlEQVR4nO2dd0AU59aHn9ldkAWCUbAhig0xejWW2(...TRUNCATED) | math_vqa | {
"ground_truth": "$$\\dfrac{1}{4}\\mathrm \\pi$$",
"style": "rule"
} | {"answer":"$$\\dfrac{1}{4}\\mathrm \\pi$$","id":"ca4f4346-11a6-4658-ab58-b92a705be2a1","index":1658,(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAHoAAABjCAYAAABUgBS3AAAzL0lEQVR4nO2dd5hV1b33P7udNpVpMDAMA6EoWBApo(...TRUNCATED) | math_vqa | {
"ground_truth": "$\\frac{1}{2}$",
"style": "rule"
} | {"answer":"$\\frac{1}{2}$","id":"c5471654-18a7-49c4-8702-386f46c2cfe8","index":2742,"question":"As s(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAEcAAABVCAIAAACUzG7YAAAcg0lEQVR4nLV8WXBdxbXo6u69zzxpnizZPpYlW1aMP(...TRUNCATED) | math_vqa | {
"ground_truth": "$13cm$",
"style": "rule"
} | {"answer":"$13cm$","id":"c46a574a-f285-446f-92c4-9258577197ff","index":3871,"question":"As shown in (...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAANYAAABxCAYAAACp+l14AAAgWklEQVR4nO2deXQUVfbHP9XdSWchC9lXA9k3SIJsI(...TRUNCATED) | math_vqa | {
"ground_truth": "5",
"style": "rule"
} | {"answer":"5","id":"becffede-d261-403a-8cd0-b8f5e5423bf6","index":1449,"question":"As shown in the f(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAfoAAAGZCAIAAADuF2TqAAEAAElEQVR4nOz915Md15Umju690+fJ4015X6gCquANC(...TRUNCATED) | math_vqa | {
"ground_truth": "4% 51",
"style": "rule"
} | {"answer":"4% 51","id":"52538d7a-f24f-4b3a-84ac-f6f038f46b4d","index":255,"question":"The stem-and-l(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAwIAAAKPCAIAAABRh+pdAAEAAElEQVR4nOz9aXMbybKuC3pE5IwZ4ExqVtUa9trHd(...TRUNCATED) | math_vqa | {
"ground_truth": "$$ 2 \\leqslant k \\leqslant 16$$",
"style": "rule"
} | {"answer":"$$ 2 \\leqslant k \\leqslant 16$$","id":"c1514916-efeb-4359-8a75-b00a251406ef","index":89(...TRUNCATED) |
FanqingM/MMK12 | [{"content":"You are a helpful assistant answering a question about an image.\n\nAlways output exact(...TRUNCATED) | [{"bytes":"iVBORw0KGgoAAAANSUhEUgAAAc8AAAGQCAYAAADIoOsJAAAk9ElEQVR4nO3dT0xT+f7/8Vd/mdzbu7KuLi5uKInOl(...TRUNCATED) | math_vqa | {
"ground_truth": "3",
"style": "rule"
} | {"answer":"3","id":"dc16525e-9522-4100-bdb1-07a74ca46b9b","index":1528,"question":"If the graph of t(...TRUNCATED) |
End of preview. Expand in Data Studio
This is the training data used in the ExpertLens paper. For further information, please refer to the project webpage, paper, and repository.
Citation
If you use ExpertLens in your research, please consider citing our work:
BibTeX:
@article{marsili2026expertlens,
title = {Harnessing Domain Specialists in Multimodal Mixture-of-Experts for Efficient Adaptation},
author = {Marsili, Damiano and Kang, Raphi and Mehta, Aditya and Perona, Pietro and Gkioxari, Georgia},
journal = {arXiv preprint arXiv:2610.02123},
year = {2026}
}
- Downloads last month
- 13
