Language Modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0.003Final Loss
15
Jun 17, 2026
0.931Validation Loss
15
Jun 8, 2026
2.86Perplexity (PPL)
15
May 29, 2026
3.372Validation Loss
15
May 20, 2026
3.338Training Loss
15
May 20, 2026
117Perplexity (PPL)
15
May 14, 2026
72Perplexity (PPL)
15
May 14, 2026
23.6PPL
15
Apr 21, 2026
71,600Throughput (tokens/s)
15
Mar 19, 2026
0.99NPM
15
Mar 5, 2026
18.7NPM
15
Mar 5, 2026
19.89NPM
15
Mar 5, 2026
5.11PPL (WikiText-2)
15
Mar 12, 2026
30,4002q1q2
15
Feb 26, 2026
10.46PPL
15
Feb 26, 2026
5.01PPL
15
Feb 26, 2026
0.62Perplexity (bits/byte)
15
Feb 26, 2026
28.18PPL (No ft)
15
Feb 26, 2026
14.09PPL
15
Feb 26, 2026
29.79Accuracy
15
Feb 26, 2026
8.94Block Efficiency
15
Apr 30, 2026
0.712BPB (val)
14
Jun 26, 2026
11.39PPL
14
Jun 2, 2026
50.5Overall Score
14
Apr 21, 2026
57.93Accuracy (Test)
14
Apr 14, 2026