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

Proof-pile

Benchmarks

Task NameDataset NameSOTA ResultTrend
Language ModelingProof-pile
Perplexity2.33
58
Language ModelingProof-pile (test)
Perplexity (2K)2.91
16
Language ModelingProof-pile 128 32k-length documents (test)
Perplexity2.28
8
Language ModelingProof-Pile
PPL2.915
6
Showing 4 of 4 rows