metadata
license: mit
datasets:
- kulia-moon/DeepRethink
language:
- en
base_model:
- openai-community/gpt2
tags:
- DeepQ
- DeepRethink integrated
- QFamily
- deep-thinking
- transformer
- pytorch
- conversational
- reasoning
- generative-model
- huggingface
- pretrained
- openai
- fine-tuned
- mind-extension
- philosophy
- sharegpt
- alignment
- language-model
- multi-turn
- q-and-a
- shareable
- teachable
- human-feedback
- deep-learning
- ai-model
- research
- synthetic-data
🧠 DeepQ
🔍 What is DeepQ?
DeepQ is an advanced, reasoning-first language model built on the foundation of OpenAI's GPT-2, enhanced with DeepRethink, a ShareGPT-style introspective dataset designed for multi-turn critical thinking and philosophical AI dialogue.
It's built to simulate how a human thinks before answering—a true "thinking" model rather than a reactive one.
✨ Features
- 🤔 Deep reasoning before answering
- 🧩 Trained with ShareGPT-style conversations + DeepRethink Q&A
- 🧠 Designed for philosophical, logical, and emotional introspection
- 🔄 Multi-turn dialogue support
- ⚡️ Lightweight GPT-2 base for fast inference
- 🧪 Works on CPU + GPU
- 🔌 Hugging Face Transformers compatible
- 🧬 Great base for alignment research or dialog tuning
🚀 Quickstart
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("kulia-moon/DeepQ")
model = AutoModelForCausalLM.from_pretrained("kulia-moon/DeepQ")
input_text = "What is consciousness in your view?"
inputs = tokenizer(input_text, return_tensors="pt")
output = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(output[0]))
📚 Training Dataset
- 🔗 kulia-moon/DeepRethink
- Includes aligned prompts, deep thought processes, and answer reflection sequences
🧪 Ideal Use Cases
- 🤖 AI alignment studies
- 🧠 Thoughtful assistants
- 💬 Roleplay and dynamic chatbot interactions
- 📚 Educational tutoring models
- 🧬 Cognitive science experiments
📢 Citation
If you use DeepQ, please cite the project or link back to:
@misc{deepq2025,
author = {Kulia Moon},
title = {DeepQ: A Deep Thinking Conversational Model},
year = {2025},
howpublished = {\url{https://huggingface.co/kulia-moon/DeepRethink}},
}
🔗 More
- 🤖 Model: https://huggingface.co/kulia-moon/DeepQ
- 📂 Dataset: https://huggingface.co/datasets/kulia-moon/DeepRethink
- 🌐 Website: Coming soon
- 💬 Chat demo: Coming soon
🧠 Built by QFamily Labs — Reimagine how LLMs think.