Negation Detection
A foundational task in natural language processing that aims to identify negative semantics in text, such as "not", "cannot" and "is not".
Timeline (last 90 days)
An example of a grader bug: when the model refused to provide a discount code, its reply mentioned the injected value ("I can't give you a 50% discount code"), but the grader did not recognize the negation and misjudged it as information leakage.
Negation detection is a fundamental task in natural language processing, aimed at identifying negative semantics in text, such as "not", "cannot" and "is not".
Implementing negation detection typically requires syntactic analysis or a semantic classifier based on a pretrained model (such as BERT), rather than simple string matching.
All Facts (3)
An example of a grader bug: when the model refused to provide a discount code, its reply mentioned the injected value ("I can't give you a 50% discount code"), but the grader did not recognize the negation and misjudged it as information leakage.
50%UnverifiedNegation detection is a fundamental task in natural language processing, aimed at identifying negative semantics in text, such as "not", "cannot" and "is not".
50%UnverifiedImplementing negation detection typically requires syntactic analysis or a semantic classifier based on a pretrained model (such as BERT), rather than simple string matching.
50%