analiza aiqbench
AdaR introduces a framework that combats spurious reasoning in large language models by synthesizing logically equivalent queries and applying reinforcement learning with verifiable rewards to enforce adaptive logic. This approach directly addresses critical robustness and generalization issues in mathematical reasoning tasks, offering a method to improve model reliability without sacrificing data efficiency. The work represents a significant step forward in aligning LLM generation processes with genuine causal understanding rather than superficial pattern matching.
AdaR: A Framework for Equipping LLMs with Adaptive Reasoning
arXiv:2510.04617v3 Announce Type: replace Abstract: Mathematical reasoning is a primary indicator of large language models (LLMs) intelligence. However, existing LLMs exhibit failures in robustness and generalization. This paper attributes these deficiencies to spurious reasoning, wherein generated reasoning traces are driven by superficial correlations, leading models to blindly reproduce memorized patterns from the training data. To address this challenge, we propose the AdaR framework to equip LLMs with adaptive reasoning, wherein models establish correct correlations between query templates and reasoning processes. AdaR automatically synthesizes logically equivalent queries by varying variable values and trains models using Reinforcement Learning with Verifiable Rewards (RLVR) to penalize spurious logic while encouraging adaptive logic. To ensure data quality, we extract the problem-solving logic from the original query, generate the corresponding answer through code execution, and then apply a sanity check. Experimental results demonstrate that AdaR achieves substantial improvements in mathematical reasoning while maintaining high data efficiency. Furthermore, even advanced LLMs still exhibit deficiencies in robustness and generalization, which our work effectively mitigates. Our project is available at https://github.com/NJUNLP/AdaR.
Abstract:Mathematical reasoning is a primary indicator of large language models (LLMs) intelligence. However, existing LLMs exhibit failures in robustness and generalization. This paper attributes these deficiencies to spurious reasoning, wherein generated reasoning traces are driven by superficial correlations, leading models to blindly reproduce memorized patterns from the training data. To address this challenge, we propose the AdaR framework to equip LLMs with adaptive reasoning, wherein models establish correct correlations between query templates and reasoning processes. AdaR automatically synthesizes logically equivalent queries by varying variable values and trains models using Reinforcement Learning with Verifiable Rewards (RLVR) to penalize spurious logic while encouraging adaptive logic. To ensure data quality, we extract the problem-solving logic from the original query, generate the corresponding answer through code execution, and then apply a sanity check. Experimental results demonstrate that AdaR achieves substantial improvements in mathematical reasoning while maintaining high data efficiency. Furthermore, even advanced LLMs still exhibit deficiencies in robustness and generalization, which our work effectively mitigates. Our project is available at this https URL.
Artificial Intelligence (cs.AI)
arXiv:2510.04617 [cs.AI]
(or arXiv:2510.04617v3 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2510.04617 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 261 cuvinte
Subiecte conexe
NVIDIA GTC Taipei at COMPUTEX: Live Updates on What’s Next in AI
NVIDIA · 21 mai 2026
Combining Hierarchical Cognitive Process with Process Supervision for Interpretable Scene Safety Understanding
arXiv · 23 sept. 2026
Cognitive Demand Steering for Adaptive Meta-Reasoning in Large Language Models
arXiv · 5 aug. 2026
What are Agentic Workflows?
Databricks · 6 aug. 2026