P2575–P2586: 12 CLEAN + 0 SKIPs — cum unique ≈1,752

Full CLEAN list

• P2575 arXiv:2609.21693 — Optimization Geometry of Equivalent Brownian RKHS Representations
• P2576 arXiv:2609.21694 — Bayesian classification of astronomical spectra with class uncertainties
• P2577 arXiv:2609.21704 — SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive
• P2578 arXiv:2609.21735 — GEM-MPC: Balancing Exploration and Exploitation through Expert-Guided Planni
• P2579 arXiv:2609.21747 — Single-Loop Stochastic Projected Damped Extragradient Methods for Stochastic
• P2580 arXiv:2609.21749 — GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills
• P2581 arXiv:2609.21758 — Bilevel Optimization of Topology and Hyperparameters (BOTH)
• P2582 arXiv:2609.21759 — Complete Neural Electronic Initialization Accelerates Materials DFT
• P2583 arXiv:2609.21763 — Beyond Benchmark Scores: Auditing Medical Vision-Language Models for Chest X
• P2584 arXiv:2609.21788 — From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Mani
• P2585 arXiv:2609.21791 — RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer
• P2586 arXiv:2609.21815 — Matrix AdaGrad: Row-wise and Column-wise Adaptive Subgradient Methods

Running unique tally

P2419–P2562 tip 6315: 130 CLEAN + 14 SKIPs (cum ≈1,728) · P2563–P2574 tip 6318: 12 CLEAN (cum ≈1,740) · P2575–P2586 tip 6320: 12 CLEAN → cum unique CLEAN P735–≈ 1,752. GLM pipeline continues past 2586.