metadata
dataset_info:
features:
- name: text
dtype: string
splits:
- name: train
num_bytes: 4814896123
num_examples: 5468096
download_size: 2179440546
dataset_size: 4814896123
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
license: odc-by
task_categories:
- text-generation
- question-answering
language:
- en
tags:
- code
- coding
- py
- python
- pretraining
- train
- fine
- clean
- filtered
- smollm
- python edu
- edu
- SLM
- smol
- programming
- github
pretty_name: Python For The Smol
size_categories:
- 1M<n<10M
SmolPyEdu
This dataset aims to be a very solid option for pre-training general-purpose or coding/programming SLMs and TLMs.
It takes Avelina/python-edu-cleaned, filters for only working scripts, removes all comments, and leaves only clean, short scripts.
It inherits the same odc-by license as the source.
Dataset Details
- Curated by: MoreThought
- Funded by: MoreThought
- Shared by: MoreThought
- License: odc-by
Dataset Sources
- Used: https://huggingface.co/datasets/Avelina/python-edu-cleaned
- Original: https://huggingface.co/datasets/Avelina/python-edu
- Source: https://huggingface.co/datasets/HuggingFaceTB/smollm-corpus
Uses
- Pre-training general-purpose or coding/programming SLMs and TLMs
- Pre-training general-purpose or coding/programming LLMs of any kind in general
- Post-training general-purpose or coding/programming SLMs and TLMs to stop using comments
Stats
| Total compressed size | Max script size | Size reduction VS used |
|---|---|---|
| 2.18 GB | 100 lines | 2.775x |