---
license: apache-2.0
task_categories:
- text-generation
dataset_info:
  features:
  - name: title
    dtype: string
  - name: content
    dtype: string
  - name: instructions
    dtype: string
  splits:
  - name: train
    num_bytes: 143897823
    num_examples: 7065
  download_size: 35205976
  dataset_size: 143897823
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
tags:
- code
---
# Scratch platformers 10k

This is a dataset featuring 10k search results for `platformer` on [Scratch](https://scratch.mit.edu/).

There are 3 fields:  
- The instrucions (how to play, ...)
- The title
- The code in the [scratchblocks](https://scratchblocks.github.io/) format

I also added `@ + the sprite name` before the blocks.