ResearchDatasetsMulti-domain composite corpusFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsBytes-per-token compressionMulti-domain composite corpus (FLORES-200 + sampled 27M corpus + HumanEval + MBPP + GSM8K)3.8Bytes/Token (English)12
Bytes-per-token compressionMulti-domain composite corpus (FLORES-200 + sampled 27M corpus + HumanEval + MBPP + GSM8K)3.8Bytes/Token (English)12