Proofpile
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Proofpile (test)
3.68Performance (8K Context)
12
ProofPile (16K)
3.24Perplexity
8
ProofPile (4K)
3.26Perplexity
8
ProofPile 100K
3.19Perplexity
6
ProofPile 32K
3.21Perplexity
6
Loading the SOTA2 catalog…