ResearchTasksFine-tuning ranking evaluationFine-tuning ranking evaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedEnglish fine-tuning corpus 60B tokensLinear-H0.817Observed Loss3May 21, 2026English fine-tuning corpus 30B tokensLinear-H0.847Observed Loss3May 21, 2026English fine-tuning corpus 3B tokensLinear-H0.904Observed Loss3May 21, 2026