Reasoning
Reasoning in LLMs refers to the model's ability to produce explicit step-by-step chain-of-thought before answering, improving performance on math, coding, and multi-step problems.
What is Reasoning?
Reasoning in LLMs refers to the model's ability to produce explicit step-by-step chain-of-thought before answering, improving performance on math, coding, and multi-step problems.
Reasoning in LLMs refers to the model's ability to produce explicit step-by-step chain-of-thought before answering, improving performance on math, coding, and multi-step problems.
Where is it used?
OpenAI o1/o3, DeepSeek-R1, and Claude 3.5 with extended thinking are reasoning models; they generate long hidden or visible reasoning traces before the final answer.
How to build it
Fine-tune a base model on chain-of-thought traces (e.g., OpenThoughts dataset) with `SFTTrainer`, or prompt with `"Let's think step by step"` to elicit reasoning at inference.