ResearchTasksBytes-per-token compressionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedMulti-domain composite corpus (FLORES-200 + sampled 27M corpus + HumanEval + MBPP + GSM8K)Airavata3.8Bytes/Token (English)12May 29, 2026
Multi-domain composite corpus (FLORES-200 + sampled 27M corpus + HumanEval + MBPP + GSM8K)Airavata3.8Bytes/Token (English)12May 29, 2026