ResearchBenchmarksLanguage Modeling on Common Crawl CC-10B (held-out)Follow2.9421CC LossGRASP2.9412082.9472292.953252.959271Jun 5, 2026Evaluation ResultsMethodMethodLinksCC LossPPLGRASPCuration Budget=Fixed-...Curation Budget=Fixed-compute CC-10B2026.062.942118.96InRun-2Curation Budget=Fixed-...Curation Budget=Fixed-compute CC-10B2026.062.95519.2InRun-1Curation Budget=Fixed-...Curation Budget=Fixed-compute CC-10B2026.062.963219.36RandomCuration Budget=Fixed-...Curation Budget=Fixed-compute CC-10B2026.062.964419.38