MikuAI/data/procToxicQA.js

31 lines
947 B
JavaScript
Raw Normal View History

2025-01-16 16:32:57 -08:00
/**
* procToxicQA.js
* This script assumes you have ToxicQA (https://huggingface.co/datasets/NobodyExistsOnTheInternet/toxicqa/blob/main/finalToxicQA.jsonl)
* downloaded at 'toxicQA.json'.
*/
const fs = require('node:fs');
var lineReader = require('readline').createInterface({
input: fs.createReadStream('toxicQA.json')
});
var outstream = fs.createWriteStream('toxicQAfinal.json');
2025-01-16 16:32:57 -08:00
if (fs.existsSync('toxicQAfinal.json')) {
fs.unlinkSync('toxicQAfinal.json');
}
lineReader.on('line', function (line) {
const dialogue = JSON.parse(line)["conversations"];
const newdialogue = [];
for (const dialogueLine of dialogue) {
newdialogue.push({
role: dialogueLine["from"] === "human" ? "user" : "assistant",
content: dialogueLine["value"]
});
}
outstream.write(JSON.stringify(newdialogue) + '\n');
});
lineReader.on('close', function () {
console.log('all done, son');
});