Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Text Generation on CNN/DailyMail (test)
Loading...
3.18
LCTG Error Rate (E)
MARKERGEN
2.0948
9.4199
16.745
24.0701
Feb 19, 2025
LCTG Error Rate (E)
Text Quality Score (S)
Token Cost (x delta)
Updated 4d ago
Evaluation Results
Method
Method
Links
LCTG Error Rate (E)
Text Quality Score (S)
Token Cost (x delta)
MARKERGEN
Backbone=Llama3.1-70B
2025.02
3.18
3.36
-
MARKERGEN
Backbone=Llama3.1-8B
2025.02
3.36
3.18
-
MARKERGEN
Backbone=Qwen2.5-32B
2025.02
4.82
3.25
-
MARKERGEN
Backbone=Qwen2.5-14B
2025.02
6.06
3.16
-
MARKERGEN
Backbone=Qwen2.5-7B
2025.02
9.92
3.07
-
Implicit
Backbone=Qwen2.5-32B
2025.02
11.05
3.21
-
Implicit
Backbone=Llama3.1-70B
2025.02
11.07
3.09
-
Implicit
Backbone=Qwen2.5-14B
2025.02
12.54
3.15
-
Implicit
Backbone=Llama3.1-8B
2025.02
15.12
3.04
-
Implicit
Backbone=Qwen2.5-7B
2025.02
30.31
3.04
-
Feedback
Search any
task
Search any
task