Language Modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
80.37BPB (%)
6
May 29, 2026
84.13BPB%
6
May 29, 2026
87.7BPB (%)
6
May 29, 2026
66.05BPB
6
May 29, 2026
0.858BPB (Bits Per Byte)
6
May 29, 2026
57.32BPB
6
May 29, 2026
7.43Perplexity (Original)
6
May 29, 2026
6.292PPL
6
May 28, 2026
2.934Loss (nats/token)
6
May 26, 2026
2.422Training Loss
6
May 20, 2026
0.582Cross-Entropy (CE)
6
May 19, 2026
26.68Average Perplexity
6
May 18, 2026
33.69Average Perplexity
6
May 18, 2026
8.09Validation Perplexity
6
May 12, 2026
18.43Accuracy
6
May 8, 2026
26.19Accuracy
6
May 8, 2026
35.08Score
6
Apr 24, 2026
8.067Perplexity (word)
6
Apr 23, 2026
79.97Perplexity
6
Apr 20, 2026
2.33Perplexity (Ctrl)
6
Apr 20, 2026
0.85Relative Loss Error
6
Apr 13, 2026
8.25Perplexity (MIMIC-III)
6
Apr 9, 2026
1.693Mean Validation Loss
6
Apr 2, 2026
0.004ΔPPL
6
Mar 31, 2026
0.001ΔPPL
6
Mar 31, 2026