Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

IMDB-review

Benchmarks

Task NameDataset NameSOTA ResultTrend
Explanation FaithfulnessIMDB Review 1,000 sentences (val)
Word Deletion Score68.9
14
Sentiment ClassificationIMDB-review heteroskedastic simulated 5%/40% noise (val)
Accuracy (Negative Class)94.3
4
Showing 2 of 2 rows