|
|
--- |
|
|
pretty_name: Simple Text Training Dataset |
|
|
license: mit |
|
|
language: |
|
|
- id |
|
|
- en |
|
|
tags: |
|
|
- nlp |
|
|
- instruction |
|
|
- simple-dataset |
|
|
task_categories: |
|
|
- text-classification |
|
|
size_categories: |
|
|
- n<1K |
|
|
--- |
|
|
|
|
|
|
|
|
# Simple Text Training Dataset |
|
|
|
|
|
Dataset ini berisi kumpulan teks pendek untuk melatih model bahasa (LLM) dalam memahami pola kalimat dasar, instruksi sederhana, dan respons percakapan. |
|
|
|
|
|
## Dataset ini cocok untuk: |
|
|
- Fine-tuning model chat sederhana |
|
|
- Latihan preprocessing NLP |
|
|
- Eksperimen tokenization |
|
|
- Project AI pemula untuk memahami alur training |
|
|
|
|
|
## Struktur Dataset |
|
|
| file | deskripsi | |
|
|
|------|-----------| |
|
|
| data.json | berisi contoh instruction–response | |
|
|
| meta.json | metadata dataset | |
|
|
|
|
|
## Format Data |
|
|
Dataset menggunakan struktur JSON seperti berikut: |
|
|
|
|
|
```json |
|
|
[ |
|
|
{ |
|
|
"instruction": "Say hello", |
|
|
"response": "Hello!" |
|
|
}, |
|
|
{ |
|
|
"instruction": "What is 2 + 2?", |
|
|
"response": "2 + 2 = 4" |
|
|
} |
|
|
] |
