File size: 1,667 Bytes
52fa950
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
---
pretty_name: Frontend React Dataset
language:
- en
task_categories:
- image-to-text
tags:
- screenshot-to-code
- react
- tsx
- ui-generation
- multimodal
size_categories:
- 1K<n<10K
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*.parquet
---

# Frontend React Dataset

This dataset contains 1,000 matched examples for training and evaluating
multimodal screenshot-to-code systems.

## Dataset structure

The dataset has one `train` split and exactly three columns:

- `screenshot`: the source webpage screenshot as an embedded PNG image
- `description`: a detailed, section-by-section visual description generated
  with Gemini 3.6 Flash
- `response`: React/TSX implementation generated with GPT-5.6 Sol

The `response` field contains extracted source code without surrounding Markdown
code fences.

## Quality checks

- All 1,000 source images were matched to their corresponding description and
  code response.
- Image SHA-256 hashes were checked while packaging the dataset.
- Every response passed static TSX syntax validation with esbuild.
- API generation status was successful for all 1,000 examples.

Static syntax validation does not guarantee pixel-perfect visual similarity or
that every example runs without its expected React dependencies.

## Intended use

This dataset is intended for research and development involving:

- screenshot-to-React generation
- multimodal code generation
- UI reconstruction
- supervised fine-tuning and evaluation

Review the source screenshots and generated code for licensing, privacy, and
redistribution requirements before making the dataset public or using it
commercially.