premiers-mots / src /experiments.js
thomwolf's picture
thomwolf HF Staff
English UI, inspectable datasets, two model sizes and labeled loss axes
249a0e9 verified
Raw History Blame Contribute Delete
1.36 kB
import { CORPUS, PREFIX } from './corpus.js';
// Original synthetic text, dedicated to CC0-1.0 alongside the story corpus.
const starts = ['lu', 'mi', 'na', 'so', 've', 'za'];
const ends = ['na', 'ri', 'lo', 'ma', 'si', 'ya'];
const names = Array.from({ length: 144 }, (_, i) =>
`name: ${starts[i % starts.length]}${ends[Math.floor(i / starts.length) % ends.length]}\n`
).join('');
export const EXPERIMENTS = [
{
id: 'pattern', title: 'Find the rhythm', tag: '01 · A PATTERN',
description: 'From noise to a regular sequence: abc, abc, abc…',
text: 'abc'.repeat(400), prefix: 'a', maxSteps: 100, temperature: 0.5,
guide: 'Watch bca appear after the prompt a. An intentionally easy repetition task.',
},
{
id: 'names', title: 'Invent names', tag: '02 · SYLLABLES',
description: 'Luna, miri, zalo… learning to combine sounds.',
text: names, prefix: 'name: ', maxSteps: 500, temperature: 0.8,
guide: 'Watch the syllables and the name format. Some names will repeat the dataset; novelty is not guaranteed.',
},
{
id: 'stories', title: 'Tell a story', tag: '03 · STORIES',
description: 'A garden, a bird, a little story to tell.',
text: CORPUS, prefix: PREFIX, maxSteps: 1000, temperature: 0.8,
guide: 'Watch words and phrases emerge. This tiny model may repeat stories and make mistakes.',
},
];