Character-level Language Modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0.94BPC
195
Feb 26, 2026
1.038BPC
128
Feb 26, 2026
1.175BPC
113
Feb 26, 2026
55.63Accuracy
48
Feb 26, 2026
1.08Bits/Char
28
Feb 26, 2026
10.08Perplexity
27
Apr 2, 2026
1.16BPC
25
Feb 26, 2026
1.04BPC
23
Jun 24, 2026
1.355Validation Loss
19
Jun 2, 2026
0.98BPC
16
Feb 26, 2026
2.3BPC
14
Jun 2, 2026
1.01BPC
13
Feb 26, 2026
59.8Accuracy
12
Apr 30, 2026
50.2Accuracy
12
Apr 30, 2026
0.72BPC
12
Feb 26, 2026
1BPC
10
Feb 26, 2026
1.24BPC
10
May 22, 2026
1.595BPC
7
Jun 18, 2026
2.23BPC
7
Jun 2, 2026
2.17Backward BPC
7
Jun 2, 2026
2.12Current BPC
7
Jun 2, 2026
2.19Forward BPC
7
Jun 2, 2026
1.441Cross-entropy Loss
6
May 13, 2026
0.186JS Divergence (Context 4)
4
Feb 26, 2026
1.222Bits per Character (BPC)
3
Jun 18, 2026