Harishapc01/quantum-qt-4q

DEMO

(https://www.youtube.com/watch?v=btVQvQEKVWM/)

⚛️ Quantum Transformer (QT-4Q) - The First Pre-trained Quantum AI Model

Quantum AI QT-4Q is a groundbreaking, quantum-inspired transformer model designed for efficient language modeling. Unlike classical transformers, this architecture utilizes quantum gates (simulated via high-fidelity linear transformations) as the core mechanism for attention and information processing.

This is the first pre-trained model released under the Quantum AI project, marking a new milestone in merging Quantum Mechanics with Generative AI.

Key Features

• Quantum-Inspired Attention: Replaces standard dot-product attention with quantum circuit-inspired gates for complex signal processing.
• State-Space Efficiency: Operates in a quantum-like state space, allowing for unique information density.
• Entropy-Based Confidence: Built-in metrics to calculate model certainty during generation using Shannon Entropy.
• BPE Tokenization: Professional Byte-Pair Encoding for superior fluency.
• KV Caching: Optimized for high-speed, low-latency inference.

Architecture Details

Parameter	Value
Model Type	Quantum Transformer (Causal LM)
Qubits	4 Qubits (Simulated)
Context Length	256 Tokens
Hidden Dimension	128
Layers	4 Quantum Blocks
Heads	4
Vocab Size	8,192

Getting Started

You can load and use this model directly using the quantum-transformer library.

Installation

pip install quantum-transformer

Inference

from quantum_transformer.model import QuantumTransformerLM

# Load the first-ever pre-trained Quantum AI model
model = QuantumTransformerLM.from_pretrained("Harishapc01/quantum-qt-4q")

# Generate text with quantum-inspired logic
output = model.generate(
    prompt="Quantum AI is the future of",
    max_new_tokens=100,
    temperature=0.8
)

print(output)

Evaluation & Confidence

One of the unique features of the QT-4Q model is its Structural Confidence Guard. During generation, it monitors the entropy of the probability distribution to ensure the output remains coherent and scientifically grounded.

License

This model is released under the MIT License.

🤝 Contact & Support

Quantum AI Team
Developer: Harisha P C
Email: reach.harishapc@gmail.com
Website: https://harishapc.com

---

Developed for the research of Quantum-Inspired Artificial Intelligence.