Large language models (LLMs) are increasingly explored for NP-hard combinatorial optimization problems, but most existing methods emphasize feasible-instance solution generation and do not explicitly address infeasibility detection. We propose an infeasibility-aware framework that combines certifiable dataset construction, supervised fine-tuning, and LLM-assisted downstream search. For the minor-embedding problem, we introduce a new mathematical programming formulation together with provable zero-phase infeasibility screening, which enables scalable construction of training instances labeled either as feasible with structured certificates or as certifiably infeasible. Using training data generated through this exact optimization pipeline, we show that an 8B-parameter LLM can be fine-tuned to jointly perform solution generation and infeasibility detection. We further utilize LLM outputs as warm starts for downstream local search, providing a practical way to accelerate optimization even when the LLM outputs are imperfect. Experiments show that our fine-tuned model improves overall accuracy by up to 30\% over GPT-5.2; meanwhile LLM-guided warm starts provide up to $2\times$ speedup compared with starting from scratch in downstream local search.
Large language models (LLMs) are increasingly explored for NP-hard combinatorial optimization problems, but most existing methods emphasize feasible-instance solution generation and do not explicitly address infeasibility detection. We propose an infeasibility-aware framework that combines certifiable dataset construction, supervised fine-tuning, and LLM-assisted downstream search. For the minor-embedding problem, we introduce a new mathematical programming formulation together with provable zero-phase infeasibility screening, which enables scalable construction of training instances labeled either as feasible with structured certificates or as certifiably infeasible. Using training data generated through this exact optimization pipeline, we show that an 8B-parameter LLM can be fine-tuned to jointly perform solution generation and infeasibility detection. We further utilize LLM outputs as warm starts for downstream local search, providing a practical way to accelerate optimization even when the LLM outputs are imperfect. Experiments show that our fine-tuned model improves overall accuracy by up to 30% over GPT-5.2; meanwhile LLM-guided warm starts provide up to 2× speedup compared with starting from scratch in downstream local search.
Scalability-constrained decision optimization requires decision systems to generate useful alternatives while simultaneously respecting computational, operational, and resource limitations. Conventional optimization approaches often depend on predefined candidate spaces, whereas generative large language models (LLMs) can construct semantically diverse decision alternatives but may produce infeasible or computationally expensive solutions. This paper proposes ScaleGen, a combinatorial generative LLM framework designed to integrate generative reasoning with explicit scalability constraints. The framework combines candidate generation, constraint representation, combinatorial evaluation, feasibility filtering, and final decision selection into a structured optimization pipeline. Its theoretical foundation is informed by research on machine learning, quantum machine learning, classification, and computational decision processes. Studies on quantum methods demonstrate the value of structured computational representations for complex learning problems (Ablayev et al., 2020a; Ablayev et al., 2020b), while quantum machine-learning applications illustrate the potential of advanced computational paradigms for classification and large decision spaces (Bhavsar et al., 2023; Yun et al., 2022). The proposed ScaleGen architecture extends these principles into an LLM-centered optimization setting by treating generated decisions as combinatorial objects subject to explicit scalability constraints. The framework emphasizes controllable generation rather than unrestricted text generation, thereby providing a conceptual bridge between generative AI and constrained optimization. The analysis indicates that ScaleGen can potentially improve candidate diversity, constraint awareness, and decision adaptability, although its effectiveness remains dependent on constraint encoding, evaluation reliability, computational overhead, and empirical validation.
Large language models (LLMs) exhibit abilities beyond natural language modelling and text generation. Recent advances in their reasoning capabilities have spurred interest in applying LLMs to complex scientific tasks requiring deep domain expertise and sophisticated reasoning. Quantum computing, as a highly specialised field with significant knowledge barriers and hardware constraints, could greatly benefit from such advancements. However, a key open question that first must be answered is: How can we develop fine-tuning pipelines that instil genuine quantum reasoning in LLMs, rather than task-specific pattern matching? We study this question through quantum circuit simulation as a training objective, where the model must predict the measurement probability distribution resulting from a sequence of quantum gate operations. We propose and compare two fine-tuning pipelines: (1) Supervised Fine-Tuning (SFT) on explicit gate-by-gate state-vector simulation traces, and (2) a two-stage SFT+Group Relative Policy Optimisation (GRPO) approach that sequentially applies SFT followed by GRPO with verifiable rewards. Our findings show that SFT achieves near-perfect in-distribution and gate-count extrapolation accuracy, significantly outperforming both the base model and the GPT-OSS-120B baseline. SFT+GRPO trades some in-distribution precision for better generalisation to larger qubit systems that SFT alone cannot handle. Both pipelines significantly outperform the baselines, demonstrating that targeted fine-tuning on explicit reasoning traces is an effective strategy for advancing quantum reasoning in LLMs.
Large language models (LLMs) have achieved remarkable outcomes in complex problems, including math, coding, and analyzing large amounts of scientific reports. Yet, few works have explored the potential of LLMs in quantum computing. The most challenging problem is to leverage LLMs to automatically generate quantum circuits at a large scale. Fundamentally, the existing pre-trained LLMs lack the knowledge of quantum circuits. In this paper, we address this challenge by fine-tuning LLMs and injecting the domain-specific knowledge of quantum computing. We describe Agent-Q, an LLM fine-tuning system to generate and optimize quantum circuits. In particular, agentQ implements the mechanisms to generate training data sets and constructs an end-to-end pipeline to fine-tune pre-trained LLMs to generate parameterized quantum circuits for various optimization problems. Agent-Q provides14,000quantum circuits covering a large spectrum of the quantum optimization landscape: 12 optimization problem instances and their optimized QAOA, VQE, and adaptive VQE circuits. Based thereon, Agent-Q fine-tunes LLMs and constructs syntactically correct parametrized quantum circuits in OpenQASM 3.0. We have evaluated the quality of the LLM-generated circuits and parameters by comparing them to the optimized expectation values and distributions. Experimental results show superior performance of Agent-Q, compared to several state-of-the-art LLMs and better parameters than random. Agent-Q can be integrated into an agentic workflow, and the generated parametrized circuits with initial parameters can be used as a starting point for further optimization, e.g., as templates in quantum machine learning and as benchmarks for compilers and hardware.
In this paper, we formulate the compression of large language models (LLMs) by optimally deleting transformer blocks (``block removal'') as a constrained binary optimization (CBO) problem that can be mapped to a physical system (Ising glass), whose energies are a strong proxy for downstream model performance. This formulation enables an efficient ranking of a large number of candidate block-removal configurations yielding many high-quality, non-trivial solutions beyond those only removing consecutive regions. Our method performs strongly in the deep compression regime, such as for 50% compression of Llama-3.3-70B-Instruct, where we achieve an almost 23 percentage point increase on the MMLU benchmark compared to other state-of-the-art (SOTA) block-removal methods. For lighter compression, it performs on par with those methods across several benchmarks for Llama-3.1-8B-Instruct, Qwen3-14B (both before and after retraining), as well as Llama-3.3-70B-Instruct. The approach is computationally efficient and requires only forward and backward passes on a calibration dataset for a few active parameters. Additionally, we demonstrate that using good heuristic solvers for the CBO problem provides solutions that perform well on downstream tasks in negligible runtime when it is unfeasible to solve the problem exactly. The method can be readily applied to any architecture. We illustrate this generality on the recent NVIDIA-Nemotron-3-Nano-30B-A3B-FP8 model, which exhibits a highly inhomogeneous and challenging block structure, and where we outperform SOTA for AIME25 and GPQA when removing either 2 attention layers or 3 mixture-of-experts layers.
Although recent advances in transfer learning have simplified training, fine-tuning big language models remains an energy-intensive and expensive process, with high hardware and energy costs that make it difficult to use and scale. While quantum machine learning (QML) presents a promising theoretical tool to overcome these bottlenecks, the practical implementation is hampered by the barren plateau problem, a phenomenon well established in literature that manifests through exponentially vanishing gradients of deep parameterized quantum circuits (PQCs) with increasing circuit depth, preventing informative parameter updates by rendering the optimization landscape flat. This study presents a hybrid quantum-classical architecture in which a PQC layer is integrated into a pretrained classical LLM backbone and fine-tuned throughout this work. This iterative refinement loop is controlled by an Expected Improvement acquisition function, which actively guides the search process through parameter regimes with obvious non-vanishing gradient variance. The study proposes that this theory-driven initialization scheme reduces the onset of barren plateaus, thereby enabling efficient and stable convergence in hybrid optimization. Experiments show that the proposed method consistently outperforms purely classical fine-tuning baselines and randomly initialized quantum baselines on representative downstream natural language processing tasks. The experimental results show that principled parameter initialization leads to concrete improvements in convergence stability, gradient trainability, and task-level performance. In summary, this research provides a scalable method for integrating near-term noisy intermediate-scale Quantum (NISQ) devices into state-of-the-art deep learning pipelines, fostering the further real-world adoption of hybrid quantum-classical systems for demanding artificial intelligence tasks.
Quantum annealing offers a promising paradigm for solving NP-hard combinatorial optimization problems, but its practical application is severely hindered by two challenges: the complex, manual process of translating problem descriptions into the requisite Quadratic Unconstrained Binary Optimization (QUBO) format and the scalability limitations of current quantum hardware. To address these obstacles, we propose a novel end-to-end framework, LLM-QUBO, that automates this entire formulation-to-solution pipeline. Our system leverages a Large Language Model (LLM) to parse natural language, automatically generating a structured mathematical representation. To overcome hardware limitations, we integrate a hybrid quantum-classical Benders' decomposition method. This approach partitions the problem, compiling the combinatorial complex master problem into a compact QUBO format, while delegating linearly structured sub-problems to classical solvers. The correctness of the generated QUBO and the scalability of the hybrid approach are validated using classical solvers, establishing a robust performance baseline and demonstrating the framework's readiness for quantum hardware. Our primary contribution is a synergistic computing paradigm that bridges classical AI and quantum computing, addressing key challenges in the practical application of optimization problem. This automated workflow significantly reduces the barrier to entry, providing a viable pathway to transform quantum devices into accessible accelerators for large-scale, real-world optimization challenges.