Technical GlossaryDeep Learning
Context Window
The maximum sequence length that a Transformer model can process in a single pass.
The context window is a critical property that defines the practical limits of Transformer-based models, especially large language models. The model can attend only to tokens within this window and must base its decisions on that visible context. In tasks such as long-document processing, multi-step reasoning, and code analysis, context size directly affects performance. It has become one of the major axes of competition in modern model development.
You Might Also Like
Explore these concepts to continue your artificial intelligence journey.
