Reasoning Switch
An annotation dimension in post-training datasets used to distinguish whether a sample contains a chain of thought or internal reasoning process. Samples with reasoning retain the reasoning steps before the final answer, while samples without reasoning retain only a concise final response.
Core Facts
Timeline (last 90 days)
The reasoning switch (reasoning on/off) is an annotation dimension introduced by many post-training datasets: samples with reasoning retain the chain of thought before the final answer, while samples without reasoning retain only a concise final answer.
For customer service, shopping guide, and notification-type reply models, what is needed is the behavioral pattern of directly giving the answer; once a large number of reasoning-on samples are mixed into the training data, the model will treat first reasoning at length and then answering as the task specification.
All Facts (2)
For customer service, shopping guide, and notification-type reply models, what is needed is the behavioral pattern of directly giving the answer; once a large number of reasoning-on samples are mixed into the training data, the model will treat first reasoning at length and then answering as the task specification.
75%UnverifiedThe reasoning switch (reasoning on/off) is an annotation dimension introduced by many post-training datasets: samples with reasoning retain the chain of thought before the final answer, while samples without reasoning retain only a concise final answer.
50%