Datasets:

Modalities:
Image
Text
Formats:
parquet
Languages:
English
ArXiv:
Libraries:
Datasets
Dask
License:
HugoLaurencon commited on
Commit
4e91fd4
1 Parent(s): 83531fd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +56 -2
README.md CHANGED
@@ -16,7 +16,61 @@ configs:
16
  data_files:
17
  - split: train
18
  path: data/train-*
 
 
 
 
 
19
  ---
20
- # Dataset Card for "image_to_website_code"
21
 
22
- [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
  data_files:
17
  - split: train
18
  path: data/train-*
19
+ license: cc-by-4.0
20
+ tags:
21
+ - code
22
+ size_categories:
23
+ - 100K<n<1M
24
  ---
25
+ # Dataset Card for TODO:FINDNAME
26
 
27
+ ## Dataset Description
28
+
29
+ This dataset comprises 823k HTML/CSS codes representing synthetically generated websites, each accompanied by a corresponding screenshot.
30
+
31
+ This dataset serves as a valuable resource for tasks such as fine-tuning vision-language models. Notably, it is used for the training of the forthcoming Idefics-2 model, aiming to augment its proficiency in generating website code based on a provided screenshot.
32
+
33
+ ## Data Fields
34
+
35
+ An example of a sample looks as follows:
36
+ ```
37
+ {
38
+ 'images': PIL.Image,
39
+ 'text': '<html>\n<style>\n{css}</style>\n{body}\n</html>',
40
+ }
41
+ ```
42
+ where `css` is the CSS code, and `body` is the body of the HTML code.
43
+ In other words, the CSS code is embedded directly within the HTML code, facilitating straightforward training of a model.
44
+
45
+ ## Data Splits
46
+
47
+ There is only one split, `train`, that contains 822,987 images and codes.
48
+
49
+ ## Dataset Creation
50
+
51
+ This dataset was created using [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) to generate random website ideas with the prompt
52
+ ```
53
+ Generate diverse website layout ideas for different companies, each with a unique design element.
54
+ Examples include: a car company site with a left column, a webpage footer with a centered logo.
55
+ Explore variations in colors, positions, and company fields.
56
+ Don't give any explanations or recognition that you have understood the request,
57
+ just give the list of 10 ideas, with a line break between each.
58
+ ```
59
+ which were then passed to [Deepseek-Coder-33b-Instruct](https://huggingface.co/deepseek-ai/deepseek-coder-33b-instruct) with the prompt
60
+ ```
61
+ Create a very SIMPLE and SHORT website with the following elements: {idea}
62
+ Be creative with the design, size, position of the elements, columns, etc...
63
+ Don't give any explanation, just the content of the HTML code `index.html` starting with `<!DOCTYPE html>`,
64
+ followed by the CSS code `styles.css` starting with `/* Global Styles */`.
65
+ Write real and short sentences for the paragraphs, don't use Lorem ipsum.
66
+ When you want to display an image, don't use <img> in the HTML, always display a colored rectangle instead.
67
+ ```
68
+ Following these steps, the HTML and CSS codes were extracted from the outputs of Deepseek-Coder and formatted into the structure `'<html>\n<style>\n{css}</style>\n{body}\n</html>'`.
69
+
70
+ ## Terms of Use
71
+
72
+ By using the dataset, you agree to comply with the original licenses of the source content as well as the dataset license (CC-BY-4.0). Additionally, if you use this dataset to train a Machine Learning model, you agree to disclose your use of the dataset when releasing the model or an ML application using the model.
73
+
74
+ ### Licensing Information
75
+
76
+ License CC-BY-4.0.