Kwai-Keye/KC-MMbench

Based on the Kuaishou short video data, we constructed 6 datasets for Vision-Language Models (VLMs) like Kwai Keye-VL-8B, Qwen2.5-VL and InternVL to evaluate performance.

Tasks

Task	Description
CPV	The task of predicting product attributes in e-commerce.
Hot_Videos_Aggregation	The task of determining whether multiple videos belong to the same topic.
Collection_Order	The task of determining the logical order between multiple videos with the same topic.
Pornographic_Comment	The task of whether short video comments contain pornographic content.
High_Like	A binary classification task to determine the rate of likes of a short video.
SPU	The task of determining whether two items are the same product in e-commerce.

Performance

Task	Qwen2.5-VL-3B	Qwen2.5-VL-7B	InternVL-3-8B	MiMo-VL-7B	Kwai Keye-VL-8B
CPV	12.39	20.08	14.95	16.66	55.13
Hot_Videos_Aggregation	42.38	46.35	52.31	49.00	54.30
Collection_Order	36.88	59.83	64.75	78.68	84.43
Pornographic_Comment	56.61	56.08	57.14	68.25	71.96
High_Like	48.85	47.94	47.03	51.14	55.25
SPU	74.09	81.34	75.64	81.86	87.05

Example of Evaluation

Here is an example of an evaluation using VLMs on our datasets. The following configuration needs to be added to the config file.

{

    "model":'...'
    "data": {
        "CPV": {
            "class": "KwaiVQADataset",
            "dataset": "CPV"
        },
        "Hot_Videos_Aggregation": {
            "class": "KwaiVQADataset",
            "dataset": "Hot_Videos_Aggregation"
        },
        "Collection_Order": {
            "class": "KwaiVQADataset",
            "dataset": "Collection_Order"
        },
        "Pornographic_Comment": {
            "class": "KwaiYORNDataset",
            "dataset": "Pornographic_Comment"
        },
        "High_like":{
            "class":"KwaiYORNDataset",
            "dataset":"High_like"
        },
        "SPU": {
            "class": "KwaiYORNDataset",
            "dataset": "SPU"
        }
    }
}