Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.04530 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912378466074624 |
|---|---|
| author | Li, Chen Luo, Yinyi Bolimera, Anudeep Ahmed, Uzair Srinivasan, Shri Kiran Gokhale, Hrishikesh Savvides, Marios |
| author_facet | Li, Chen Luo, Yinyi Bolimera, Anudeep Ahmed, Uzair Srinivasan, Shri Kiran Gokhale, Hrishikesh Savvides, Marios |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_04530 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SOLAR: Scalable Optimization of Large-scale Architecture for Reasoning Li, Chen Luo, Yinyi Bolimera, Anudeep Ahmed, Uzair Srinivasan, Shri Kiran Gokhale, Hrishikesh Savvides, Marios Artificial Intelligence 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. |
| title | SOLAR: Scalable Optimization of Large-scale Architecture for Reasoning |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2503.04530 |