chain-of-thought prompting
A strategy used in prompting models to encourage them to produce intermediate reasoning steps, thereby enhancing logical coherence and accuracy in outputs.
- Evaluating the Inductive Abilities of Large Language Models: Why Chain-of-Thought Reasoning Sometimes Hurts More Than Helps
- Language Models Can Predict Their Own Behavior
- MMPerspective: Do MLLMs Understand Perspective? A Comprehensive Benchmark for Perspective Perception, Reasoning, and Robustness
- Pause Tokens Strictly Increase the Expressivity of Constant-Depth Transformers
- Scaling Code-Assisted Chain-of-Thoughts and Instructions for Model Reasoning
- Unlabeled Data Can Provably Enhance In-Context Learning of Transformers
- Visual Structures Help Visual Reasoning: Addressing the Binding Problem in LVLMs
- When Thinking Fails: The Pitfalls of Reasoning for Instruction-Following in LLMs