QuLoC: Photonic Quantum-Assisted Low-Rank LLM Compression
Abstract
As LLMs grow in size, compression becomes increasingly important for efficient deployment. SVD-based low-rank compression reduces parameter counts but can degrade downstream performance. To improve performance after compression, we introduce QuLoC, a photonic quantum-assisted LLM compression algorithm that uses quantum circuit outputs to gate the retained low-rank components during training. Model performance is recovered through local functional reconstruction followed by end-to-end knowledge distillation. After training, the gating coefficients are absorbed into the low-rank factors, allowing the compressed model to run on classical hardware without executing quantum circuits during inference. Experiments on Qwen3.5-4B show that QuLoC achieves a 9.73\% relative improvement in average accuracy over state-of-the-art baselines across multiple downstream tasks, demonstrating its effectiveness. We further evaluate QuLoC on LLaMA-7B at different parameter compression ratios. It achieves comparable or higher average downstream accuracy than state-of-the-art baselines, supporting its applicability to a larger model across different compression settings. These results motivate further exploration of photonic quantum-assisted compression for larger models and more complex agentic tasks.
Related
- Quantum Simulation of Non-Markovian Open Quantum Systems via Low-Rank Kernel Compression and Coupling Dilution
- Optimal low-rank compression of quantum dynamics
- Accelerated S-NFC for Million-Chaff RCS Computation Using Low-Rank Compression of Concatenated Block Rows
- Scattering-based detection of fungal spores on a photonic integrated platform: an ASCA-assisted proof-of-concept study
- Learning photonic quantum states
BibTeX
@misc{ni2026quloc,
title = {{QuLoC: Photonic Quantum-Assisted Low-Rank LLM Compression}},
author = {Xiao-Hui Ni and Yu-Han Yao and Yu-Ze Zhu and Hang Song and Xiang Zhao and Lin Yang and Xian-Min Jin},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2609.40146},
eprint = {2609.40146},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.40146},
}
RIS
TY - UNPB
TI - QuLoC: Photonic Quantum-Assisted Low-Rank LLM Compression
AU - Xiao-Hui Ni
AU - Yu-Han Yao
AU - Yu-Ze Zhu
AU - Hang Song
AU - Xiang Zhao
AU - Lin Yang
AU - Xian-Min Jin
PY - 2026
DA - 2026/10/02
PB - arXiv
DO - 10.48550/arxiv.2609.40146
UR - https://arxiv.org/abs/2609.40146
AB - As LLMs grow in size, compression becomes increasingly important for efficient deployment. SVD-based low-rank compression reduces parameter counts but can degrade downstream performance. To improve performance after compression, we introduce QuLoC, a photonic quantum-assisted LLM compression algorithm that uses quantum circuit outputs to gate the retained low-rank components during training. Model performance is recovered through local functional reconstruction followed by end-to-end knowledge distillation. After training, the gating coefficients are absorbed into the low-rank factors, allowing the compressed model to run on classical hardware without executing quantum circuits during inference. Experiments on Qwen3.5-4B show that QuLoC achieves a 9.73\% relative improvement in average accuracy over state-of-the-art baselines across multiple downstream tasks, demonstrating its effectiveness. We further evaluate QuLoC on LLaMA-7B at different parameter compression ratios. It achieves comparable or higher average downstream accuracy than state-of-the-art baselines, supporting its applicability to a larger model across different compression settings. These results motivate further exploration of photonic quantum-assisted compression for larger models and more complex agentic tasks.
ER -