Datasets:

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. 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.

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