aagoluoglu/text-to-svg

---
license: cc-by-nc-sa-4.0
task_categories:
- text-to-image
language:
- en
tags:
- svg
- code-generation
- text-to-svg
pretty_name: NYU DL Spring 2026 – SVG Generation from Text Prompts
---

# DL Spring 2026 – SVG Generation from Text Prompts - Kaggle Competition Dataset

## Dataset Description

Training and test data from the NYU Tandon Deep Learning (ECE-GY 7123)
Spring 2026 Kaggle competition: **[DL Spring 2026 – SVG Generation from Text Prompts](https://www.kaggle.com/competitions/dl-spring-2026-svg-generation/overview)**.

This dataset was provided by the course instructors as competition data.
It is redistributed here for reproducibility of the associated course project.

## Source

- **Course**: NYU Tandon Deep Learning (ECE-GY 7123) Spring 2026
- **Competition**: [Kaggle: DL Spring 2026 – SVG Generation from Text Prompts](https://www.kaggle.com/competitions/dl-spring-2026-svg-generation/overview)

## Files

| File | Rows | Description |
|------|------|-------------|
| `train.csv` | 50,000 | `id`, `prompt`, `svg` — ground-truth pairs for training |
| `test.csv` | 1,000 | `id`, `prompt` — prompts to generate SVGs for |
| `train_filtered.csv` | varies | Filtered subset used for fine-tuning (generated by `01_data_exploration.ipynb`) |

## Dataset Structure
```python
# train.csv columns:
# id — unique row identifier (UUID)
# prompt — natural language description (~20 words average)
# svg — ground-truth SVG code (256x256 canvas)

# test.csv columns:
# id — unique row identifier
# prompt — natural language description
```

## License

[CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/)

This data may be shared and adapted for **non-commercial purposes**,
with attribution, and any derivatives must use the same license.

## Citation

If you use this dataset, please credit the source competition:
```
NYU Tandon Deep Learning Spring 2026 Kaggle Competition
"DL Spring 2026 – SVG Generation from Text Prompts"
https://www.kaggle.com/competitions/dl-spring-2026-svg-generation/overview
```