Hybrid Quantum Transformer for Language Generation

TL;DR

This paper introduces HyQuT, the first hybrid quantum-classical large language model that integrates quantum circuits into the Transformer architecture, demonstrating feasibility for large-scale natural language generation.

Contribution

It presents a novel hybrid quantum-classical LLM architecture that incorporates variational quantum circuits into Transformers at large scales.

Findings

Quantum circuits can replace about 10% of classical parameters in a 150M-parameter model.

The hybrid model achieves comparable convergence stability and generation quality to classical models.

Early evidence of quantum computing's potential in large-scale language generation.

Abstract

Although quantum computing has been increasingly applied to replace classical computation, most existing quantum or hybrid models remain confined to simple tasks, with no successful application to large-scale natural language generation to date. In this work, we present the first hybrid quantum-classical large language model (LLM) for natural language generation, HyQuT, capable of performing coherent and context-aware dialogue. The proposed architecture integrates variational quantum circuits (VQCs) into the Transformer framework at both 8M and 150M parameter scales. Experimental results show that a minimal number of qubits (10 qubits with 80 quantum gates) can replace about 10% of the classical parameters in the 150M-parameter model, while achieving comparable convergence stability and generation quality. This study provides an early demonstration of the feasibility of integrating quantum computing to large-scale generative language models.