SOLAR: Scalable Optimization of Large-scale Architecture for Reasoning

Large Language Models excel in reasoning yet often rely on Chain-of-Thought prompts, limiting performance on tasks demanding more nuanced topological structures. We present SOLAR (Scalable Optimization of Large-scale Architecture for Reasoning), a framework that dynamically optimizes Chain-of-Thought (CoT), Tree-of-Thought (ToT), and Graph-of-Thought (GoT) topologies to boost accuracy and efficiency. Our Topological-Annotation-Generation (TAG) system automates dataset creation, annotation, and difficulty segmentation, leading to stronger post training and test-time performance. We also propose Topological-Scaling, a curriculum-learning-based approach that adaptively combines post training and inference scaling to each task. On MATH and GSM8K, SOLAR delivers notable gains: +5% accuracy with Topological Tuning, +9% with Topological Rewarding, and +10.02% with Hybrid Scaling, while reducing response length by over 5%, lowering inference latency. To further enhance efficiency, we introduce a multi-task Topological Reward Model (M-TRM) that selects both the optimal reasoning topology and final answer in a single pass, eliminating multiple single-task TRMs. Remarkably, M-TRM also surpasses all single-task TRMs, improving accuracy by +10% and rank correlation by +9%. Overall, SOLAR establishes a new benchmark for scalable, high-precision LLM reasoning and introduces a fully automated, dynamic topology competition mechanism.

@article{li2025_2503.04530,
  title={ SOLAR: Scalable Optimization of Large-scale Architecture for Reasoning },
  author={ Chen Li and Yinyi Luo and Anudeep Bolimera and Uzair Ahmed and Shri Kiran Srinivasan and Hrishikesh Gokhale and Marios Savvides },
  journal={arXiv preprint arXiv:2503.04530},
  year={ 2025 }
}