metadata
license: cc0-1.0
language:
- en
- fr
- tr
tags:
- art
- emoji
- brainrot
pretty_name: Youtube shorts comments dataset
size_categories:
- 100K<n<1M
YouTube Shorts Comments Dataset
Acollection of 1.4M+ comments scraped from YouTube Shorts - a mix of brainrot, emoji spam, and English, Turkish, French, and unidentifiable langauges. Perfect for NLP experiments, meme research, or just laughing at internet absurdity.
- Inspired from htmx (Idk how to call him lmao)
- Fine-tuned distilgpt2 on this dataset
Facts 🤔
- I calculated that average amount of emojis in a YouTube short comment is 4.59
- Sources: This dataset, Python script
Dataset Details 🔍
- Rows: 1,475,500 comments
- Original Size: 105 MB (raw)
- Parquet Size: 41 MB (optimized)
- Sample Content: Emoji storms, random links, broken English etc.
Fields 📋
Single column:text (string) - The raw comment text with emojis/links
Suggested Uses 🤖
- Emoji frequency analysis
- Short-form text classification
- Internet culture studies
- Training a "cursed" language model
- Measuring how many 😂 fit in one comment
License 📜
CC0 1.0 Universal (Public Domain)
Do whatever you want - no attribution required
Shoutouts 🔊
- Subscribe to FaceDev (not sponsored, just cool)
- Join my Discord Server: PingVortex Discord
"12 millions view 😂😂😂 good god 🤦♂️" - Anonymous YouTube Shorts Kid