Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Unconditional Text Generation on OpenWebText (held-out segments)
Loading...
1
MAUVE
*Data
-0.01712
0.24694
0.511
0.77506
Nov 7, 2025
MAUVE
LLaMA2 Score
LLaMA3 Score
GPT-2 Score
Entropy
Updated 1mo ago
Evaluation Results
Method
Method
Links
MAUVE
LLaMA2 Score
LLaMA3 Score
GPT-2 Score
Entropy
*Data
Sampling steps=512
2025.11
1
7
9.4
14.8
5.44
AR
Sampling steps=512
2025.11
0.855
10.97
15.12
12.07
5.21
KLASS
Sampling steps=512
2025.11
0.179
26.94
49.19
45.5
5.43
MDLM
Sampling steps=512
2025.11
0.115
30.88
54.15
51.78
5.46
SEDD
Sampling steps=512
2025.11
0.037
53.09
109.6
105.4
5.62
D3PM
Sampling steps=512
2025.11
0.022
41.82
72.85
76.7
5.4
Feedback
Search any
task
Search any
task