On an idiom completion dataset, the negative log likelihood of GPT-2 XL's top token, measured across residual stream layers, predicts whether the completion is correct with an AUC of 0.92.
Srivastava, and Jürgen Schmidhuber
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Uncovering Uncertainty in Transformer Inference
On an idiom completion dataset, the negative log likelihood of GPT-2 XL's top token, measured across residual stream layers, predicts whether the completion is correct with an AUC of 0.92.