squ11z1/Hypnos-i1-8B

Hugging Face
TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Nov 22, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

Hypnos i1-8B by squ11z1 is a specialized 8 billion parameter reasoning model based on Nous Hermes 3 (Llama 3.1 8B) designed for complex logic, chain-of-thought reasoning, and mathematical problem-solving. It features a unique "Quantum Noise Injection" training methodology, utilizing real entropy data from IBM Quantum Heron processors as a stochastic regularizer. This quantum-informed approach aims to enhance the model's creativity and reduce deterministic patterns, making it suitable for tasks requiring deep, multi-step logical inference.

Loading preview...

Hypnos i1-8B: Quantum-Informed Reasoning Model

Hypnos i1-8B is a specialized 8 billion parameter model built upon the Nous Hermes 3 (Llama 3.1 8B) architecture, focusing on advanced reasoning, complex logic, and mathematical problem-solving. Its core differentiator is a novel training methodology involving "Quantum Noise Injection." This process incorporates real entropy data generated by IBM Quantum Heron processors (133/156-qubit architecture) during supervised fine-tuning, acting as a stochastic regularizer to improve creativity and prevent mode collapse.

Key Capabilities

  • S-Tier Reasoning: Excels in logic and math, rivaling larger 70B class models in specific, narrow tasks like multi-step logic puzzles and causal inference.
  • Quantum-Informed: The first known LLM fine-tuned with raw measurement data from 100+ qubit GHZ states from IBM's latest quantum hardware.
  • Uncensored & Compliant: Based on Nous Hermes 3, it follows instructions without refusal while maintaining safety.
  • Deep Thinker: Optimized for long-context reasoning (4096+ tokens), often "thinking out loud" for higher accuracy on complex queries.

Good For

  • Complex Logic & Math: Ideal for tasks requiring intricate reasoning and problem-solving.
  • Creative Generation: The quantum-informed training aims to reduce deterministic patterns, potentially enhancing creative outputs.
  • Edge & Prototyping: Its 8B size makes it suitable for deployment on laptops and for rapid experimental development.

Finetunes
2 models