nelsonmaligro's picture
Update README.md
a00ca1a verified
|
Raw
History Blame Contribute Delete
3.03 kB
metadata
license: cc-by-4.0
task_categories:
  - text-generation
language:
  - en
pretty_name: Universal Chat SFT Dataset
size_categories:
  - 1K<n<10K
tags:
  - chat
  - sft
  - instruction-tuning
  - conversational-ai
  - llm
  - dialogue
  - synthetic
viewer: true

Universal-Chat-SFT-Dataset

A large-scale multi-turn conversational dataset designed for supervised fine-tuning (SFT) of modern Large Language Models (LLMs).

The dataset follows the OpenAI-style chat format with structured messages, making it directly compatible with most modern LLM training frameworks including Hugging Face Transformers, TRL, Axolotl, Unsloth, LlamaFactory, and custom fine-tuning pipelines.


Features

✅ Multi-turn conversations

✅ OpenAI-compatible chat message format

✅ System, User and Assistant roles

✅ Ready for supervised fine-tuning


Supported Models

This dataset can be used for supervised fine-tuning of chat models including:

  • Llama 3
  • Qwen 2 / 2.5 / 3
  • Mistral
  • Gemma
  • Phi
  • DeepSeek
  • Falcon
  • GPT-style decoder-only models
  • Any instruction-tuned causal language model

Dataset Schema

Each row contains one complete conversation.

Column Type Description
text list Complete conversation

Conversation Format

{
  "text": [
    {
      "role": "user",
      "content": "..."
    },
    {
      "role": "assistant",
      "content": "..."
    }
  ]
}

Loading the Dataset

from datasets import load_dataset

dataset = load_dataset(
    "Srijan-Chakraborty/Universal-Chat-SFT-Dataset"
)

print(dataset)

Intended Use

This dataset is intended for:

  • Supervised Fine-Tuning (SFT)
  • Instruction Tuning
  • Conversational AI
  • Chatbot Training
  • Dialogue Modeling
  • LLM Alignment
  • Benchmarking
  • Research
  • Education

Compatibility

The dataset is directly compatible with:

  • Hugging Face Transformers
  • Hugging Face Datasets
  • TRL
  • Axolotl
  • Unsloth
  • LlamaFactory
  • vLLM preprocessing
  • Custom PyTorch pipelines

Citation

If you use this dataset in your research, please cite:

@misc{universal_chat_sft_dataset,
  author = {Srijan Chakraborty},
  title = {Universal-Chat-SFT-Dataset},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/Srijan-Chakraborty/Universal-Chat-SFT-Dataset}
}

Author

Srijan Chakraborty

GitHub: https://github.com/SrijanChakraborty2003

Hugging Face: https://huggingface.co/Srijan-Chakraborty


License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Users are free to share and adapt the dataset for any purpose, including commercial use, provided appropriate attribution is given.

For the full license text, see:

https://creativecommons.org/licenses/by/4.0/