ResearchDatasetsProof-pileProof-pileFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsLanguage ModelingProof-pile2.33Perplexity130Language ModelingProof-pile (test)2.91Perplexity (2K)16Language ModelingProof-pile 128 32k-length documents (test)2.28Perplexity8Language ModelingProof-Pile2.915PPL6Self-attention inverse temperature scaling analysisProof-Pile-2 (PP2)7.2Tie Percentage2