rudyon/lainchan-alpaca

---
license: mit
task_categories:
- text-generation
- question-answering
language:
- en
tags:
- imageboard
- programming
pretty_name: Lainchan Alpaca
size_categories:
- 10K<n<100K
---

# lainchan-alpaca

Alpaca style instruct dataset scraped from various boards on [lainchan](https://lainchan.org). The data was gathered from threads and replies on the boards /λ/, /Ω/, /lit/, /hum/, /inter/ and /Δ/.

This dataset was scraped with the intention to finetune a model with a "lainanon" like personality and vibe. The resulting model is [rudyon/lainanon-0.6b](https://huggingface.co/rudyon/lainanon-0.6b).

## format

The dataset follows the Alpaca format with the following columns:
- `instruction`: the question or prompt from a lainchan post
- `input`: additional context (usually empty)
- `output`: the response from a lainchan reply
- `board`: the board the post was scraped from