HyperQ Adds Quantum Circuits to 1.1B Diffusion LM, Beats LoRA

HyperQ bolts token-conditioned quantum residual branches onto a frozen 1.1-billion-parameter masked-diffusion language model, then evaluates the circuits classically. The trick is that "the required expectation values have an exact classical expression whose evaluation cost grows linearly with the qubit count," which lets the authors sweep circuits from 16 to 64 qubits inside a full-size backbone without touching real quantum hardware.

The paper, Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models by Xiaoqiang Wang, Mengyang Xiong, Jun Dai and Bang Liu, reports that widening the circuits raises the average downstream benchmark score from 47.65 to 54.30. At 64 qubits, HyperQ "exceeds the backbone and its low-rank-adapted counterpart by 4.71 and 3.67 points, respectively." It gets there while being "fine-tuned on 20,000 prompt-response pairs, compared with 200,000 for the classical baselines."

The abstract names no specific benchmarks, publishes no per-task numbers, and does not report training or inference wall-clock against LoRA. The authors frame the result as support for "token-conditioned circuit emission as a tractable architectural approach to quantum-augmented language modelling."