Unconditional Text Generation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
1.07Gen. PPL
227
Jun 24, 2026
36.11Perplexity
64
Feb 26, 2026
27.35Generation Perplexity
30
May 20, 2026
86.8MAUVE
27
May 18, 2026
4.29Entropy
24
May 12, 2026
692.3LLAMA2 Score
21
Feb 26, 2026
90.1Mauve Score
18
Apr 3, 2026
8.1Entropy (Ent)
13
May 11, 2026
47.46BLEU
11
Feb 26, 2026
5.24LM Score
8
Feb 26, 2026
0.512Human Score
8
Feb 26, 2026
14.4c-PPL
7
Jul 2, 2026
0Repetition Percentage
7
Jul 2, 2026
0.893MAUVE
7
Jun 16, 2026
2.9Training Duration (days)
6
May 11, 2026
1MAUVE
6
Mar 9, 2026
1MAUVE
6
Mar 9, 2026
31.82Average Perplexity
6
Mar 4, 2026
3.41Clarity
4
Mar 24, 2026
—
—Primary metric
0
Feb 26, 2026