Leveraging Self-Motivated Learning to Enhance Language Model Reasoning Capabilities
By leveraging the inherent preference that a rationale leading to the correct answer is superior to one leading to an incorrect answer, this work proposes a self-motivated learning framework to enhance the reasoning capabilities of language models without relying on large models or extensive manual annotations.
Corex: Enhancing Complex Reasoning through Multi-Model Collaboration
Corex, a suite of collaborative reasoning strategies, transforms large language models into autonomous agents that can work together to enhance complex reasoning capabilities through Discuss, Review, and Retrieve modes.
Explaining the Reasoning Ability of Pretrained Large Language Models through Template-Content Structure
The template-content structure, where the generated sequence is split into a fixed template and flexible content, is the key to explaining how pretrained large language models can solve complex reasoning tasks with limited training data.
Can Small Language Models Enhance Reasoning in Large Language Models?
Leveraging a lightweight language model to guide a large language model in reasoning tasks can improve the quality of generated rationales and enhance overall task performance.
Evaluating the Reasoning Capabilities of Large Language Models: Beyond Accuracy
Large language models have demonstrated impressive performance on reasoning tasks, but their depth of reasoning abilities remains uncertain. This survey provides a comprehensive review of studies that go beyond task accuracy, offering deeper insights into the models' reasoning processes.
