Mini toolsDefinition
Capabilities of large AI models not explicitly trained for, appearing abruptly as model scale increases.
Detailed Explanation
Capabilities or behaviors exhibited by large AI models that were not explicitly programmed or directly trained for, and which often appear somewhat abruptly as model scale increases.
Use Cases
Understanding scaling effects in AI, researching capabilities of foundation models, predicting future AI advancements, studying phenomena like few-shot learning or chain-of-thought reasoning.
