Temporal Reasoning
The ability of a model to understand, compute and manipulate temporal information, including parsing temporal expressions, computing time differences and determining the order of events.
Timeline (last 90 days)
In NLP, temporal reasoning generally refers to a model's ability to understand, compute and manipulate temporal information, such as parsing temporal expressions, computing time differences and determining the order of events.
The study argues that large models' performance on temporal reasoning benchmarks depends largely on memorized temporal common sense rather than transferable reasoning ability.
Model evaluation experiments based on cognitive demands show that quantitative reasoning and logical reasoning are not the main drivers of performance on temporal reasoning benchmarks; knowledge is the core driver.
Transfer analysis across two benchmark families shows that the so-called 'coherent temporal reasoning ability' lacks evidentiary support.
All Facts (4)
In NLP, temporal reasoning generally refers to a model's ability to understand, compute and manipulate temporal information, such as parsing temporal expressions, computing time differences and determining the order of events.
50%UnverifiedThe study argues that large models' performance on temporal reasoning benchmarks depends largely on memorized temporal common sense rather than transferable reasoning ability.
50%UnverifiedModel evaluation experiments based on cognitive demands show that quantitative reasoning and logical reasoning are not the main drivers of performance on temporal reasoning benchmarks; knowledge is the core driver.
50%UnverifiedTransfer analysis across two benchmark families shows that the so-called 'coherent temporal reasoning ability' lacks evidentiary support.
50%