sijanpaudel/nepali-recipes-qwen-processed

Nepali Recipes for Qwen Fine-tuning

Dataset Description

This dataset contains 1227 Nepali recipes formatted for fine-tuning Qwen models using ChatML format.

• —Train Split: 900 recipes
• —Test Split: 327 recipes
• —Language: Nepali (ne)
• —Format: Qwen ChatML
• —Base Model: Qwen/Qwen2-1.5B

Dataset Structure

Data Fields

• —text: Full ChatML formatted prompt with answer (for training)
• —test_text: ChatML prompt without answer (for inference)
• —name: Recipe name in Nepali
• —ingredients: List of ingredients
• —instructions: List of cooking instructions
• —token_count: Number of tokens in the prompt

Token Statistics

• —Minimum: 359 tokens
• —Maximum: 2043 tokens
• —Average: 943 tokens

Example

from datasets import load_dataset Load dataset

ds = load_dataset('sijanpaudel/nepali-recipes-qwen-processed') View first training example

print(ds['train']['text']) print(ds['train']['name']) print(ds['train']['ingredients'])

text

Additional Files

This repository also contains:

• —train.pkl: Pickled Recipe objects for training
• —test.pkl: Pickled Recipe objects for testing
• —train_prompts.jsonl: Training prompts in JSONL format
• —test_prompts.jsonl: Test prompts in JSONL format
• —metadata.json: Dataset statistics and information

Loading Pickle Files

from huggingfacehub import hfhub_download import pickle Download pickle file

trainfile = hfhubdownload( repoid='sijanpaudel/nepali-recipes-qwen-processed', filename='train.pkl', repo_type='dataset' ) Load Recipe objects

with open(train_file, 'rb') as f: train = pickle.load(f) Access properties

print(train.prompt) print(train.ingredients) print(train.instructions)

text

Usage for Fine-tuning

from datasets import load_dataset Load dataset

dataset = load_dataset('sijanpaudel/nepali-recipes-qwen-processed') Use for training

train_texts = dataset['train']['text'] Use for evaluation

testtexts = dataset['test']['testtext']

text

Source

Original dataset: sijanpaudel/nepali-recipes

License

MIT License

Citation

If you use this dataset, please cite:

@dataset{nepalirecipesqwen_2025, author = {Sijan Paudel}, title = {Nepali Recipes for Qwen Fine-tuning}, year = {2025}, publisher = {Hugging Face}, url = {https://huggingface.co/datasets/sijanpaudel/nepali-recipes-qwen-processed} }

text