List your company

Intrinsic Tokenization Evaluation on ClimbMix (held-out)

99Vocabulary UtilizationDet
20.27240.71161.1581.589May 21, 2026

Evaluation results

MethodLinks
DetVocabulary Size=8k2026.05990.005610.617.021,141.34.0962.30.0014
BiasVocabulary Size=8k2026.05990.005410.697.11,1713.9364.60.0013
DetVocabulary Size=16k2026.0598.80.012310.96.831,833.84.46570.0015
BiasVocabulary Size=16k2026.0598.80.01210.956.881,860.94.3658.30.0015
IntVocabulary Size=16k2026.0598.80.008210.186.991,033.64.0363.60.0014
BPEVocabulary Size=8k2026.0598.60.005510.617.091,121.53.9963.80.0014
IntVocabulary Size=8k2026.0598.60.00319.316.9492.33.3975.60.0011
BPEVocabulary Size=16k2026.0598.10.01210.96.881,780.44.3858.10.0015
IntVocabulary Size=32k2026.0593.40.019510.826.791,975.14.55560.0015
BiasVocabulary Size=32k2026.0592.60.024211.066.722,609.14.6854.40.0016
DetVocabulary Size=32k2026.0592.40.024411.036.72,586.14.7253.90.0016
BPEVocabulary Size=32k2026.0592.10.02411.046.732,516.84.6754.50.0016
IntVocabulary Size=64k2026.0573.80.033610.986.662,760.24.8530.0016
BiasVocabulary Size=64k2026.0571.50.038811.076.633,226.24.8552.40.0017
DetVocabulary Size=64k2026.0571.20.038911.056.623,210.44.8852.10.0017
BPEVocabulary Size=64k2026.0570.70.038411.076.643,1384.8552.40.0017
IntVocabulary Size=128k2026.0545.80.046311.036.583,410.64.9651.30.0017
BiasVocabulary Size=128k2026.05440.04911.046.583,566.74.9751.20.0017
DetVocabulary Size=128k2026.0543.80.048911.036.583,550.84.9851.10.0017
BPEVocabulary Size=128k2026.0543.30.048111.056.593,504.54.9651.30.0017
IntVocabulary Size=256k2026.0525.20.0514116.563,5635.0150.70.0017
BiasVocabulary Size=256k2026.0523.70.053411.016.553,671.15.0250.60.0017
DetVocabulary Size=256k2026.0523.60.053211.016.553,660.95.0350.60.0017
BPEVocabulary Size=256k2026.0523.30.052511.026.563,635.45.0250.70.0017