shubhamprakash108/simple-human-chat-10k

Simple Human Chat 10K

A 10,000-row human conversation dataset for training or fine-tuning small chatbot models, especially lightweight models around 33M parameters.

Each row contains one user prompt and one assistant-style response. The dataset focuses on simple everyday conversation such as greetings, casual replies, polite questions, short support responses, and basic human-like dialogue.

Dataset Details

This dataset contains simple prompt-response pairs for conversational AI training.

Column	Description
prompt	User message
response	Assistant reply

Rows: 10,000 Format: CSV Split: train

How to Use

First install the Hugging Face datasets library in your terminal or notebook:

pip install datasets

Then run this Python code:

from datasets import load_dataset

dataset = load_dataset("shubhamprakash108/simple-human-chat-10k")

print(dataset)
print(dataset["train"][0])

def format_example(example):
    return {
        "text": f"User: {example['prompt']}\nAssistant: {example['response']}"
    }

formatted_dataset = dataset["train"].map(format_example)

print(formatted_dataset[0]["text"])

Example output:

DatasetDict({
    train: Dataset({
        features: ['prompt', 'response'],
        num_rows: 10000
    })
})

{'prompt': 'Hi! How are you today?', 'response': "Hi! I'm doing well today, thanks. How are you?"}

User: Hi! How are you today?
Assistant: Hi! I'm doing well today, thanks. How are you?

Intended Use

This dataset is intended for basic chatbot training, small language model fine-tuning, conversational AI experiments, learning simple human-like response patterns, and testing prompt-response training pipelines.

It is useful for teaching a model simple conversation behavior like greetings, polite replies, casual questions, and short assistant responses.

Not Intended For

This dataset is not designed for factual knowledge training, advanced reasoning, medical advice, legal advice, financial advice, safety-critical chatbot behavior, or production assistants without additional safety training.

Example Row

prompt	response
Hi! How are you today?	Hi! I'm doing well today, thanks. How are you?

Notes

The dataset was generated with deterministic templates and variation rules to avoid repeated prompt-response pairs.

If you find this dataset useful, please like it on Hugging Face.