sujitpandey/phone-specs-qa-alpaca

Phone Specifications QA Dataset (Alpaca Format)

Dataset Description

This dataset contains 222 question-answer pairs based on phone specifications for the Xiaomi Redmi A4 5G. The data is formatted in Alpaca format, making it suitable for fine-tuning large language models (LLMs) using frameworks like Unsloth.

Dataset Summary

• Total Examples: 222
• Format: Alpaca (instruction, input, output)
• Task: Question Answering on Phone Specifications
• Phone Model: Xiaomi Redmi A4 5G
• Language: English

Purpose

This dataset is designed to fine-tune LLMs (like Granite) to answer questions about phone specifications without requiring the full specification context in every prompt. After fine-tuning, the model should internalize phone specification knowledge and respond accurately to queries.

Dataset Structure

Data Format

Each entry contains three fields following the Alpaca format:

• instruction: A question about the phone specifications
• input: The complete phone specification text
• output: The answer to the question

Example Entry

{'instruction': 'What is the brand and model of Xiaomi Redmi A4 5G?', 'output': 'The phone is Xiaomi Redmi A4 5G.'}

Files

• training_dataset.jsonl - JSONL format (one JSON object per line) - Main training file

Coverage

The dataset covers all major aspects of phone specifications:

• Basic Information (brand, model)
• Physical Specifications (dimensions, weight, colors)
• SIM Card Details
• Network Support (2G, 3G, 4G, 5G)
• Operating System
• Processor and Chipset
• Memory (RAM)
• Storage
• Display Specifications
• Sensors
• Camera System (rear and front)
• Audio Features
• Connectivity (Wi-Fi, Bluetooth, USB, GPS)
• Battery Specifications
• Supported Media Formats

Usage

Using with Unsloth

from datasets import load_dataset
from unsloth import FastLanguageModel

# Load dataset
dataset = load_dataset("sujitpandey/phone-specs-qa-alpaca")

# Load model
model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "ibm-granite/granite-3.0-1b-a400m-instruct",
    max_seq_length = 2048,
    dtype = None,
    load_in_4bit = True,
)

# Fine-tune using the dataset
# ... (add your fine-tuning code)

Using with Hugging Face Datasets

from datasets import load_dataset

# Load dataset
dataset = load_dataset("sujitpandey/phone-specs-qa-alpaca")

# Access training examples
train_data = dataset["train"]

# Iterate through examples
for example in train_data:
    print(f"Question: {example['instruction']}")
    print(f"Answer: {example['output']}")

Direct JSONL File Loading

import json

# Load JSONL file directly
with open("training_dataset.jsonl", "r", encoding="utf-8") as f:
    data = [json.loads(line) for line in f]

Dataset Creation

The dataset was generated programmatically using generate_dataset.py which:

1. Reads complete phone specifications from a text file
2. Creates diverse question-answer pairs covering all specification aspects
3. Formats data in Alpaca format (instruction, input, output)
4. Exports to JSONL format for efficient training

Use Cases

• Fine-tuning LLMs for phone specification Q&A
• Training models to reduce prompt context size
• Creating domain-specific chatbots for mobile devices
• Educational purposes for LLM fine-tuning
• Benchmark for phone specification understanding

Limitations

• Dataset is based on a single phone model (Xiaomi Redmi A4 5G)
• For production use, expand to multiple phone models
• Answers are concise and factual (not conversational)

Citation

If you use this dataset, please cite:

@dataset{phone_specs_qa_alpaca,
  title={Phone Specifications QA Dataset (Alpaca Format)},
  author={Your Name},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/sujitpandey/phone-specs-qa-alpaca}
}

License

MIT License - Free to use for commercial and non-commercial purposes.