PreprintNot peer reviewedArtificial IntelligenceResearchers report vision models can boost weaker audio and video MLLMs via mergingOriginal title: Strong Helps Weak: Directional Cross-Modal Alignment Transfer in Multi-modal LLMsA preprint reports that merging a well-aligned vision model into audio or video multi-modal LLMs improves their target-modality performance without extra training.Hoigi Seo et al. · Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingResearchers report GNN-based post-selection for quantum error correction codesOriginal title: Graph Neural Post-selection for Quantum Error CorrectionA preprint reports graph neural networks that flag likely decoder failures from syndrome data alone, cutting logical error rates without extra decoder runs.Conor Carty et al. · Added 24 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceA preprint reports a jailbreak defence that extracts intent from LLM outputsOriginal title: Reactivating Alignment: Defending LLMs from Jailbreaks via Intention-Aware Input-Output MatchingA preprint describes SENTINEL, a generation-time defence that matches LLM input and output text to spot and block jailbreak attempts.Luoyu Chen et al. · Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingResearchers report running several quantum circuit compilations together to boost shot throughputOriginal title: Compiling Together: High-Throughput Distributed Quantum Computing via Multi-CompilationA preprint proposes running multiple equivalent circuit compilations together in distributed quantum computing to raise shot throughput and fidelity.Yipei Liu et al. · Added 24 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceA preprint reframes LLM agent design using deep learning concepts like layers and NASOriginal title: Building LLM Agent Systems the Deep Learning Way: From Modular Design to Architecture SearchA preprint proposes building LLM agent systems as modular architectures, borrowing ideas like backpropagation and neural architecture search.Tao Feng et al. · Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingPreprint reports exponential quantum space edge in randomly ordered data streamsOriginal title: Exponential quantum space advantage in random data streamsA preprint reports quantum algorithms can use exponentially less memory than classical ones on two streaming problems, even with randomly ordered data.Adam Bouland et al. · Added 24 Oct 2026
PreprintNot peer reviewedFrom Dell TechnologiesArtificial IntelligenceResearchers propose analytical model for choosing LLM inference parallelism strategiesOriginal title: Characterizing Parallelism Strategies in LLM Inference: Fundamental Compute-Communication Trade-offsA preprint presents a compute-communication model explaining when tensor versus pipeline parallelism works best for LLM inference.Added 24 Oct 2026
PreprintNot peer reviewedFrom IBM ResearchQuantum ComputingA preprint proposes variance-reduced gradient methods to cut quantum optimisation circuit costsOriginal title: Breaking the Shot-Noise Barrier: Cached Recycled Variance-Reduced Gradients for Quantum OptimizationResearchers report a caching-based gradient technique for quantum optimisation that trims circuit use but often trails a simpler baseline.Added 24 Oct 2026
PreprintNot peer reviewedArtificial IntelligencePreprint proposes training LLMs to internalise a safe persona, not just refusal phrasesOriginal title: Beyond Refusal Patterns: Safe-Role Internalization for Robust and Generalizable LLM Safety AlignmentResearchers report a fine-tuning method that teaches models safety-oriented behaviour rather than refusal patterns, improving robustness and reducing over-refusal.Jinghao Pang et al. · Added 24 Oct 2026
PreprintNot peer reviewedFrom QudeLeap ResearchQuantum ComputingPreprint describes optimal-query circuit for multivariate quantum polynomialsOriginal title: Multivariate quantum signal processing with optimal query complexityA preprint presents a quantum circuit implementing multivariate trigonometric polynomials with query cost matching a proven lower bound.Added 24 Oct 2026
PreprintNot peer reviewedFrom XanaduQuantum ComputingPreprint describes a denser QROM encoding that cuts quantum lookup costs furtherOriginal title: Halving the cost of QROM again via dense encodingA preprint reports a new QROM encoding that uses both Pauli bases of dirty qubits to cut quantum lookup costs beyond earlier methods.Added 24 Oct 2026
PreprintNot peer reviewedFrom SGIT AI LabArtificial IntelligencePreprint reports difficulty-aware evolution strategy improves LLM fine-tuning over standard ESOriginal title: DART-ES: Difficulty-Aware Reweighting and Targeted Replay for Fine-Tuning LLMs with Evolution StrategiesA preprint describes DART-ES, a method that reweights training problems by difficulty to improve evolution-strategy fine-tuning of language models.Added 24 Oct 2026
PreprintNot peer reviewedFrom TencentQuantum ComputingResearchers report a unitary synthesis method optimal in both size and depth-ancilla tradeoffOriginal title: Exact Unitary Synthesis: Simultaneously Optimal in Size and Depth-Ancilla TradeoffA preprint describes quantum circuits that reach optimal gate count while also hitting the best possible depth for any ancilla budget.Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingResearchers report simplified epitaxial NbN/AlN/NbN junctions for superconducting qubitsOriginal title: Sidewall spacer passivated epitaxial NbN/AlN/NbN Josephson junctions for superconducting qubit applicationsA preprint describes a sidewall-spacer fabrication process for epitaxial NbN/AlN/NbN Josephson junctions that the authors say preserves electrical quality while cutting complexity.Koki Honda et al. · Added 24 Oct 2026
PreprintNot peer reviewedFrom Alibaba GroupArtificial IntelligenceA preprint reports FP4 rollout training method for MoE language models matching BF16 performanceOriginal title: TRACE: Rollout-Guided Quantization-Aware Training for FP4 Reinforcement Learning of MoE Language ModelsResearchers propose TRACE, an FP4 quantization scheme for RL training of MoE language models, reporting BF16-level performance with faster rollout.Added 24 Oct 2026
PreprintNot peer reviewedArtificial IntelligenceFramework aligns federated mixture-of-experts fine-tuning across heterogeneous mobile devicesOriginal title: Federated Mixture-of-Experts Alignment on Mobile Edge Networks under Data HeterogeneityA preprint proposes FedAlign-MoE, which aligns gating and expert behaviour across devices to improve federated LLM fine-tuning on non-IID data.Zihan Fang et al. · Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingPreprint describes gradient-based optimisation framework for superconducting qubit designOriginal title: A General Framework for Gradient-Based Optimization of Superconducting Quantum Circuits using Qubit Discovery as a Case StudyA preprint presents an automatic-differentiation framework for optimising superconducting quantum circuits, applied to qubit discovery.Taha Rajabzadeh et al. · Added 24 Oct 2026
PreprintNot peer reviewedFrom NEC Labs AmericaArtificial IntelligencePreprint proposes workflow-affinity embeddings to tailor agentic LLM workflows per queryOriginal title: RA-MoWE: Workflow-Affinity Embeddings for Query Clustering and Agentic Workflow GenerationA preprint reports that clustering queries by how they respond to reference workflows can boost task scores while cutting LLM calls.Added 24 Oct 2026
PreprintNot peer reviewedFrom Google Quantum AIQuantum ComputingResearchers report shallower circuits for block-encoding matrix polynomials on quantum hardwareOriginal title: Practical block encodings of matrix polynomials that can also be trivially controlledA preprint describes circuit designs that make block-encoding matrix polynomials and controlling them far cheaper for near-term quantum devices.Added 24 Oct 2026
PreprintNot peer reviewedQuantum ComputingPreprint describes nucleation method for preparing quantum states on digital computersOriginal title: Digital quantum state preparation by nucleationResearchers report a nucleation-inspired technique for preparing lattice-model ground states that can beat a leading state-of-the-art approach in 2D tests.Jean Paul Sadia et al. · Added 24 Oct 2026
