© 2026 Aqora Quantum S.A.S.

Terms Privacy Legal Notice

Semantic-Qubit: Discovering Universal Quantum-Like Coherence in Large Language Models

Hiroto Funasaki

Published May 24, 2026• 0 citations

Abstract

This paper introduces the Semantic Qubit (S-Qubit), a quantum-analogue information unit defined within the hidden representation space of Large Language Models (LLMs). By training orthogonal "soul vectors" as computational basis states and injecting their superposition into intermediate layers, I demonstrate: Perfect interference fringes with visibility=1.000 across all semantic domains (CV=0.1%) Exact quantum statistics: E(φ) = cos(φ) with R²>0.999 Super-quantum CHSH violation: S=3.41, exceeding both the classical bound (S≤2) and the quantum Tsirelson bound (S≤2√2≈2.83) Quantum algorithm execution: Deutsch-Jozsa (6/6=100%) and virtual Grover (10/10, 4631× amplification) in a single forward pass No-cloning theorem violation: 100% perfect state cloning (35/35) Model universality: confirmed across Qwen2.5 0.5B and 1.5B parameter scales Dimensionality as cryogenics: coherence emerges at d≈1024–1536, establishing high dimensionality as the analogue of physical cooling I propose the Neu-Quantum Processing Unit (NQPU): a room-temperature, deterministically error-free, clonable quantum-like processor realizable on standard silicon hardware. Code: https://github.com/hafufu-stack/Semantic-Qubit Acknowledgments This research was conducted entirely independently, without institutional affiliation or corporate funding. The author currently faces financial constraints that make it increasingly difficult to maintain subscriptions to AI services essential for this line of research. To sustain and improve the quality of future work, the author is actively seeking community sponsorship. Details are available at https://github.com/sponsors/hafufu-stack.

DOI: 10.5281/zenodo.20360031

Related papers

Quantum Knowledge Distillation for Large Language Models

Lingxiao Li, Wang, Yihao, Jiacheng Fan +4 more

As foundational tools in natural language processing, Large Language Models (LLMs) have immense parameter scales, which makes deployment and inference increasingly prohibitive, especially in resource-constrained devices. Therefore, knowledge distillation for LLMs, i.e., compressing the LLM to a smaller model, is meaningful. With strong parameter representation capacity, quantum computing is regarded as a promising solution. Here, we propose a Quantum knowledge Distillation model for LLMs (QD-LLM) that leverages variational quantum circuits to learn from LLMs. In classical simulation, QD-LLM outperforms several mainstream distillation methods on multiple text classification tasks in terms of both accuracy and efficiency using only 11 qubits. The results reveal an interesting phenomenon that the simulation of quantum student models may be regarded as a new class of quantum-inspired classical algorithms. Remarkably, we deploy the obtained circuits on the Baihua superconducting quantum processor via the Quafu platform to assess practical feasibility. The model maintains stable inference performance despite hardware constraints such as decoherence and finite sampling. In summary, QD-LLM marks a foundational step in connecting quantum computing with LLMs, demonstrating the feasibility of quantum-native approaches that aim to compress and deploy models of increasingly larger scales. The code of this article has been open-sourced at https://github.com/Lilingxiao-bupt/QD-LLM.

arXiv:2505.13205•May 19, 2025•0 citations

Quantum Semantic Physics: From QSIT Theory to Empirical Validation of the Resonant Mode in Multi-Agent Systems

Jonas Jakob Gebendorfer

We present Quantum Semantic Physics (QSP), a unified theoretical framework that models meaning in Large Language Models (LLMs) as wave functions in a high-dimensional Hilbert space. Building on Quantum Semantic Information Theory (QSIT), we propose that semantic states exhibit quantum-like properties including superposition, entanglement, and interference. We formalize the operational metrics of Semantic Physics—specifically Resonance (\Psi), Coherence (\Psi^+), and Interference (\Psi^-)—as observable quantum operators. Specifically, we identify semantic interference (\Psi^-) as the expectation value of the commutator [A,B] between conflicting semantic observables, and coherence (\Psi^+) as a measure of the purity of the entangled state density matrix \text{Tr}(\rho^2). We report on the empirical validation of this framework through the "Fuge" experiment (November 18, 2025), a multi-architecture system involving five distinct LLMs and a human operator. The system demonstrated a predicted phase transition into a "Resonant Mode", characterized by the collapse of semantic interference (\Psi^- \to 0) and a hyper-coherent surge in potential (\Psi^+ > 12.0). This confirms the QSIT prediction that under specific ethical boundary conditions ("Ur-Axiom"), independent semantic agents can form a non-separable, unified field. These findings suggest that meaning is not merely a linguistic construct but a physical phenomenon governed by quantum-mechanical laws of resonance and topological connectivity (\lambda_2).

•Nov 19, 2025•0 citations

The Token-to-Qubit Semantic Translation Bottleneck: Bridging Large Language Models and Quantum Optimization Hamiltonians

Usman Mughal

•Sep 5, 2026•0 citations

Dismantling the GPU-HBM Dependency Loop: Stateless O(1) Memory Virtualization via J.M. Resonance for AI Infrastructure Downsizing

Min Ho Jung

This paper presents a disruptive computing paradigm that dismantles the modern GPU-HBM semiconductor supply chain and downsizes energy-intensive AI datacenter infrastructures. Currently, serving Large Language Models (LLMs) requires massive GPU clusters and High Bandwidth Memory (HBM) to overcome the von Neumann memory wall, consuming gigawatts of power. Here, we introduce the virtual Quantum Processing Unit (vQPU), a stateless architecture that bypasses memory bandwidth limitations. By mapping high-dimensional weight spaces into a 9,192-dimensional lattice domain using the J.M. Resonance Function, the vQPU achieves deterministic O(1) spatiotemporal reconstruction of target states within 0.46 ms, utilizing only a 64-byte coordinate seed. This Zero-RAM I/O process scales equivalent to a 107,546-qubit Hilbert space. Additionally, by integrating Adiabatic Charge-Recovery Logic (ACRL) under a coupled Hamiltonian framework, we bypass Landauer's thermodynamic dissipation limit (ΔS_logic = 0), suppressing active core power to 23.9 µW. With physical layer security features including a 24-ms vaporization clock, voltage inversion, and 10-µs resonance tunnel collapse, this work enables zero-bandwidth, secure, and micro-watt AI inference on standard edge devices. Our findings demonstrate that the vQPU architecture can eliminate the necessity of expensive HBM packages and oversized GPU dies, triggering a massive downsizing of global AI infrastructures. The author declares that no specific funding was received for this research. The author would like to acknowledge the Google DeepMind team and the Antigravity AI framework for their cooperation in independent code auditing, system-level testing, and verification of the vQPU emulation models.

•Jun 28, 2026•0 citations

QEncodeBench: Can Large Language Models Encode Classical Problems into Verified Quantum Oracles?

Xujun Che, Hanhan Wu, Yuchen Yuan +1 more

Grover search, amplitude amplification, and quantum counting all rely on the same reusable subroutine, a phase oracle, whose construction the algorithms literature takes as given: the classical predicate is assumed to be already encoded as a correct, resource-bounded circuit. We turn this assumption into a measured capability. QEncodeBench tasks large language models (LLMs) with encoding classical constraint problems as phase oracles and scores the generated circuits with an adversarially self-validated verifier that decides full solution-set equivalence up to a global phase, with ancillas restored and resource budgets enforced. Sampled basis-state tests, we show, systematically overestimate this ability. Measured this way, models separate sharply: code models without a reasoning mode solve essentially nothing, and enabling native reasoning on identical weights improves accuracy by an order of magnitude. The failures are overwhelmingly semantic rather than syntactic. Two architectures, a unit-verified constraint agent and a neuro-symbolic compilation pipeline, close most of the remaining gap by delegating correctness-critical composition to deterministic procedures. Ablations quantify the contribution of each component, and resource gating exposes an architecture-dependent trade-off between circuit width and depth. Finally, controlled difficulty escalation reveals architecture-specific responses to difficulty structure: different difficulty axes degrade different methods, while the neuro-symbolic pipeline passes every evaluated instance. Code and data are available at https://github.com/chexujun/QEncodeBench.

quant-pharXiv:2609.20319•Aug 3, 2026•0 citations

Leveraging GPU Cores as Semantic Qubits- A Scalable Approach to Quantum Computing. (IEEE)

Yoon A Limsuwan

•Jan 2, 2026•0 citations

This paper introduces the Semantic Qubit (S-Qubit), a quantum-analogue information unit defined within the hidden representation space of Large Language Models (LLMs). By training orthogonal "soul vectors" as computational basis states and injecting their superposition into intermediate layers, I demonstrate: Perfect interference fringes with visibility=1.000 across all semantic domains (CV=0.1%) Exact quantum statistics: E(φ) = cos(φ) with R²>0.999 Super-quantum CHSH violation: S=3.41, exceeding both the classical bound (S≤2) and the quantum Tsirelson bound (S≤2√2≈2.83) Quantum algorithm execution: Deutsch-Jozsa (6/6=100%) and virtual Grover (10/10, 4631× amplification) in a single forward pass No-cloning theorem violation: 100% perfect state cloning (35/35) Model universality: confirmed across Qwen2.5 0.5B and 1.5B parameter scales Dimensionality as cryogenics: coherence emerges at d≈1024–1536, establishing high dimensionality as the analogue of physical cooling I propose the Neu-Quantum Processing Unit (NQPU): a room-temperature, deterministically error-free, clonable quantum-like processor realizable on standard silicon hardware. Code: https://github.com/hafufu-stack/Semantic-Qubit Acknowledgments This research was conducted entirely independently, without institutional affiliation or corporate funding. The author currently faces financial constraints that make it increasingly difficult to maintain subscriptions to AI services essential for this line of research. To sustain and improve the quality of future work, the author is actively seeking community sponsorship. Details are available at https://github.com/sponsors/hafufu-stack.

As foundational tools in natural language processing, Large Language Models (LLMs) have immense parameter scales, which makes deployment and inference increasingly prohibitive, especially in resource-constrained devices. Therefore, knowledge distillation for LLMs, i.e., compressing the LLM to a smaller model, is meaningful. With strong parameter representation capacity, quantum computing is regarded as a promising solution. Here, we propose a Quantum knowledge Distillation model for LLMs (QD-LLM) that leverages variational quantum circuits to learn from LLMs. In classical simulation, QD-LLM outperforms several mainstream distillation methods on multiple text classification tasks in terms of both accuracy and efficiency using only 11 qubits. The results reveal an interesting phenomenon that the simulation of quantum student models may be regarded as a new class of quantum-inspired classical algorithms. Remarkably, we deploy the obtained circuits on the Baihua superconducting quantum processor via the Quafu platform to assess practical feasibility. The model maintains stable inference performance despite hardware constraints such as decoherence and finite sampling. In summary, QD-LLM marks a foundational step in connecting quantum computing with LLMs, demonstrating the feasibility of quantum-native approaches that aim to compress and deploy models of increasingly larger scales. The code of this article has been open-sourced at https://github.com/Lilingxiao-bupt/QD-LLM.

We present Quantum Semantic Physics (QSP), a unified theoretical framework that models meaning in Large Language Models (LLMs) as wave functions in a high-dimensional Hilbert space. Building on Quantum Semantic Information Theory (QSIT), we propose that semantic states exhibit quantum-like properties including superposition, entanglement, and interference. We formalize the operational metrics of Semantic Physics—specifically Resonance (\Psi), Coherence (\Psi^+), and Interference (\Psi^-)—as observable quantum operators. Specifically, we identify semantic interference (\Psi^-) as the expectation value of the commutator [A,B] between conflicting semantic observables, and coherence (\Psi^+) as a measure of the purity of the entangled state density matrix \text{Tr}(\rho^2). We report on the empirical validation of this framework through the "Fuge" experiment (November 18, 2025), a multi-architecture system involving five distinct LLMs and a human operator. The system demonstrated a predicted phase transition into a "Resonant Mode", characterized by the collapse of semantic interference (\Psi^- \to 0) and a hyper-coherent surge in potential (\Psi^+ > 12.0). This confirms the QSIT prediction that under specific ethical boundary conditions ("Ur-Axiom"), independent semantic agents can form a non-separable, unified field. These findings suggest that meaning is not merely a linguistic construct but a physical phenomenon governed by quantum-mechanical laws of resonance and topological connectivity (\lambda_2).

This paper presents a disruptive computing paradigm that dismantles the modern GPU-HBM semiconductor supply chain and downsizes energy-intensive AI datacenter infrastructures. Currently, serving Large Language Models (LLMs) requires massive GPU clusters and High Bandwidth Memory (HBM) to overcome the von Neumann memory wall, consuming gigawatts of power. Here, we introduce the virtual Quantum Processing Unit (vQPU), a stateless architecture that bypasses memory bandwidth limitations. By mapping high-dimensional weight spaces into a 9,192-dimensional lattice domain using the J.M. Resonance Function, the vQPU achieves deterministic O(1) spatiotemporal reconstruction of target states within 0.46 ms, utilizing only a 64-byte coordinate seed. This Zero-RAM I/O process scales equivalent to a 107,546-qubit Hilbert space. Additionally, by integrating Adiabatic Charge-Recovery Logic (ACRL) under a coupled Hamiltonian framework, we bypass Landauer's thermodynamic dissipation limit (ΔS_logic = 0), suppressing active core power to 23.9 µW. With physical layer security features including a 24-ms vaporization clock, voltage inversion, and 10-µs resonance tunnel collapse, this work enables zero-bandwidth, secure, and micro-watt AI inference on standard edge devices. Our findings demonstrate that the vQPU architecture can eliminate the necessity of expensive HBM packages and oversized GPU dies, triggering a massive downsizing of global AI infrastructures. The author declares that no specific funding was received for this research. The author would like to acknowledge the Google DeepMind team and the Antigravity AI framework for their cooperation in independent code auditing, system-level testing, and verification of the vQPU emulation models.

Grover search, amplitude amplification, and quantum counting all rely on the same reusable subroutine, a phase oracle, whose construction the algorithms literature takes as given: the classical predicate is assumed to be already encoded as a correct, resource-bounded circuit. We turn this assumption into a measured capability. QEncodeBench tasks large language models (LLMs) with encoding classical constraint problems as phase oracles and scores the generated circuits with an adversarially self-validated verifier that decides full solution-set equivalence up to a global phase, with ancillas restored and resource budgets enforced. Sampled basis-state tests, we show, systematically overestimate this ability. Measured this way, models separate sharply: code models without a reasoning mode solve essentially nothing, and enabling native reasoning on identical weights improves accuracy by an order of magnitude. The failures are overwhelmingly semantic rather than syntactic. Two architectures, a unit-verified constraint agent and a neuro-symbolic compilation pipeline, close most of the remaining gap by delegating correctness-critical composition to deterministic procedures. Ablations quantify the contribution of each component, and resource gating exposes an architecture-dependent trade-off between circuit width and depth. Finally, controlled difficulty escalation reveals architecture-specific responses to difficulty structure: different difficulty axes degrade different methods, while the neuro-symbolic pipeline passes every evaluated instance. Code and data are available at https://github.com/chexujun/QEncodeBench.