Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,61 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: cc-by-sa-4.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-sa-4.0
|
| 3 |
+
language:
|
| 4 |
+
- zh
|
| 5 |
+
---
|
| 6 |
+
Based on the [Kuaishou](https://www.kuaishou.com/) short video data, we constructed 6 datasets for Vision-Language Models (VLMs) like **Keye**, **Qwen2.5-VL** and **InternVL** to evaluate performance.
|
| 7 |
+
|
| 8 |
+
## Tasks
|
| 9 |
+
| Task | Description |
|
| 10 |
+
| -------------- | --------------------------------------------------------------------------- |
|
| 11 |
+
| CPV | The task of predicting product attributes in e-commerce. |
|
| 12 |
+
| Video_Topic | The task of determining whether multiple videos belong to the same topic. |
|
| 13 |
+
| Video_Order | The task of determining the logical order between multiple videos. |
|
| 14 |
+
| PornComment | The task of whether short video comments contain pornographic content. |
|
| 15 |
+
| High_Like | A binary classification task to determine the rate of likes of a short video. |
|
| 16 |
+
| SPU | The task of determining whether two items are the same product in e-commerce. |
|
| 17 |
+
|
| 18 |
+
## Performance
|
| 19 |
+
| Task | Qwen2.5-VL-3B | Qwen2.5-VL-7B | InternVL-3-8B | MiMo-VL | Keye |
|
| 20 |
+
| -------------- | ------------- | ------------- | ------------- | ------- | ---- |
|
| 21 |
+
| CPV | 12.80 | 20.10 | 15.00 | 17.10 | 55.13 |
|
| 22 |
+
| Video_Topic | 43.90 | 46.95 | 51.21 | 49.39 | 54.30 |
|
| 23 |
+
| Video_Order | 36.80 | 58.40 | 64.80 | 78.40 | 84.43 |
|
| 24 |
+
| PornComment | 57.10 | 56.50 | 57.60 | 68.60 | 71.96 |
|
| 25 |
+
| High_Like | 48.20 | 48.70 | 47.80 | 50.40 | 55.25 |
|
| 26 |
+
| SPU | 74.10 | 81.30 | 75.60 | 81.90 | 87.05 |
|
| 27 |
+
|
| 28 |
+
## Example of Evaluation
|
| 29 |
+
|
| 30 |
+
Here is an example of an evaluation using VLMs on our datasets. The following configuration needs to be added to the config file.
|
| 31 |
+
```python
|
| 32 |
+
{
|
| 33 |
+
|
| 34 |
+
"model":'...'
|
| 35 |
+
"data": {
|
| 36 |
+
"CPV": {
|
| 37 |
+
"class": "KwaiVQADataset",
|
| 38 |
+
"dataset": "CPV"
|
| 39 |
+
},
|
| 40 |
+
"Video_Topic": {
|
| 41 |
+
"class": "KwaiVQADataset",
|
| 42 |
+
"dataset": "Video_Topic"
|
| 43 |
+
},
|
| 44 |
+
"Video_Order": {
|
| 45 |
+
"class": "KwaiVQADataset",
|
| 46 |
+
"dataset": "Video_Order"
|
| 47 |
+
},
|
| 48 |
+
"PornComment": {
|
| 49 |
+
"class": "KwaiYORNDataset",
|
| 50 |
+
"dataset": "PornComment"
|
| 51 |
+
},
|
| 52 |
+
"High_like":{
|
| 53 |
+
"class":"KwaiYORNDataset",
|
| 54 |
+
"dataset":"High_like"
|
| 55 |
+
},
|
| 56 |
+
"SPU": {
|
| 57 |
+
"class": "KwaiYORNDataset",
|
| 58 |
+
"dataset": "SPU"
|
| 59 |
+
}
|
| 60 |
+
}
|
| 61 |
+
}
|