A new 313-sentence peer-review toxicity benchmark shows GPT-4 with detailed instructions reaches a Cohen's Kappa of 0.56 with human judges.
In Proceedings of the 2013 conference on empirical methods in natural language processing, 1631–1642
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Benchmark on Peer Review Toxic Detection: A Challenging Task with a New Dataset
A new 313-sentence peer-review toxicity benchmark shows GPT-4 with detailed instructions reaches a Cohen's Kappa of 0.56 with human judges.