Update README.md
Browse files
README.md
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
---
|
|
|
2 |
dataset_info:
|
3 |
features:
|
4 |
- name: SAMPLE_ID
|
@@ -27,7 +28,7 @@ dataset_info:
|
|
27 |
- name: train
|
28 |
num_bytes: 28506248899
|
29 |
num_examples: 107166507
|
30 |
-
download_size:
|
31 |
dataset_size: 28506248899
|
32 |
configs:
|
33 |
- config_name: default
|
@@ -35,3 +36,38 @@ configs:
|
|
35 |
- split: train
|
36 |
path: data/train-*
|
37 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
license: cc-by-4.0
|
3 |
dataset_info:
|
4 |
features:
|
5 |
- name: SAMPLE_ID
|
|
|
28 |
- name: train
|
29 |
num_bytes: 28506248899
|
30 |
num_examples: 107166507
|
31 |
+
download_size: 16353125308
|
32 |
dataset_size: 28506248899
|
33 |
configs:
|
34 |
- config_name: default
|
|
|
36 |
- split: train
|
37 |
path: data/train-*
|
38 |
---
|
39 |
+
# 100M Text Debiased Subset from LAION 2B
|
40 |
+
|
41 |
+
- Captions in LAION-2B have a significant bias towards describing visual text content embedded in the images.
|
42 |
+
- Released CLIP models have strong text spotting bias in almost every style of web images, resulting in the CLIP-filtering datasets inherently biased towards visual text dominant data.
|
43 |
+
- CLIP models easily learn text spotting capacity from parrot captions while failing to connect the vision-language semantics, just like a text spotting parrot.
|
44 |
+
|
45 |
+
For more details, please see our [paper](https://arxiv.org/abs/2312.14232).
|
46 |
+
|
47 |
+
## Filtering Details
|
48 |
+
|
49 |
+
We provide an alternative solution by releasing a less biased filtered LAION-2B 100M(107,166,507) subset.
|
50 |
+
|
51 |
+
We construct a less biased 100M subset from the LAION-2B subset with Empty OCR results, CLIP score > 0.3, and Aesthetics score > 4.5.
|
52 |
+
|
53 |
+
We add the ase_scores and K-means labels (4000 total) for each image-text pair.
|
54 |
+
|
55 |
+
*We also released the dataset on [OpenDataLab](https://openxlab.org.cn/datasets/opendatalab-linyiqi/LAION-text-debiased-100M).*
|
56 |
+
|
57 |
+
The pre-trained CLIP model is released on [github](https://github.com/opendatalab/CLIP-Parrot-Bias).
|
58 |
+
|
59 |
+
## Reference
|
60 |
+
```
|
61 |
+
@article{lin2023parrot,
|
62 |
+
title={Parrot Captions Teach CLIP to Spot Text},
|
63 |
+
author={Yiqi Lin and Conghui He and Alex Jinpeng Wang and Bin Wang and Weijia Li and Mike Zheng Shou},
|
64 |
+
journal={arXiv preprint arXiv:2312.14232},
|
65 |
+
year={2023}
|
66 |
+
}
|
67 |
+
@misc{conghui2022opendatalab,
|
68 |
+
author={He, Conghui and Li, Wei and Jin, Zhenjiang and Wang, Bin and Xu, Chao and Lin, Dahua},
|
69 |
+
title={OpenDataLab: Empowering General Artificial Intelligence with Open Datasets},
|
70 |
+
howpublished = {\url{https://opendatalab.com}},
|
71 |
+
year={2022}
|
72 |
+
}
|
73 |
+
```
|