pretty_name: Pashto User CPT Chunks
tags:
- pashto
- cpt-format
- user-side
- llm-training
- text-generation
- dialogue
- talanda-tm
- nasimjp
- spinzar-enterprises
language:
- ps
license: cc-by-4.0
size_categories:
- 100K<n<500K
task_categories:
- text-generation
dataset_info:
corporate_owner: Spinzar Enterprises Inc. (Yūgen Kaisha), Japan
division: Pashto AI Research & Development
creator: Nassim Nasibullah
mission: >
Establishing the first dedicated corporate R&D track for Pashto-language AI,
focusing on large-scale dataset creation, model training, and foundational
LLM infrastructure.
💎 Pashto User CPT Chunks
Dataset ID: nassimjp/user_cpt_chunks
Modality: Text
Format: JSONL
Size: 39 MB
Rows: 262,586
Split: train
📘 Description
This dataset contains Pashto conversational training data in CPT (Conversational Pre‑Training) format.
Each record is a single JSON object:
{"text": "<pashto_line>"}
The corpus represents user‑side dialogue turns extracted from multi‑chunk translation pipelines used in Pashto LLM development. It is optimized for instruction‑tuning, dialogue modeling, and language understanding downstream tasks.
🧩 Source
Generated from the user_chunks translation batches of the Railway‑TM pipeline. Converted using the convert_to_pashto_cpt() script to ensure strict, clean JSONL compliance.
🏗️ Structure
| Field | Type | Description |
|---|---|---|
text |
string | Pashto sentence or user query |
🪶 Tags
pashto · llm-training · cpt-format · dialogue · user-side · text-generation · language-modeling · railway-tm · nasimjp · ai-dataset
⚙️ Usage
Example loading snippet via Hugging Face datasets:
from datasets import load_dataset
ds = load_dataset("nassimjp/user_cpt_chunks", split="train")
print(ds[0])
📜 License
This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.
🌍 Author & Affiliation
Nassim JP — Pashto AI Research Investor
Part of the Talanda‑TM Pashto LLM Corpus Series.