EpochLex entry

Long-Context Evaluation

long KON-tekst ih-val-yoo-AY-shun

Evaluation designed to measure how well a model uses, retrieves from, reasons over, or summarizes information across very long input contexts rather than merely accepting a large context window.