Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Sequential Model Editing on CounterFact T = 10000
Loading...
86
Efficacy
BetaEdit
-3.128
20.011
43.15
66.289
May 10, 2026
Efficacy
Generality
Specificity
Updated 22d ago
Evaluation Results
Method
Method
Links
Efficacy
Generality
Specificity
BetaEdit
Backbone=GPT-J-6B
2026.05
86
41.8
4.1
AlphaEdit
Backbone=GPT-J-6B
2026.05
82.6
40.3
3.9
BetaEdit
Backbone=Qwen3-4B-Inst...
2026.05
78.5
54.8
3.8
BetaEdit
Backbone=LLaMA3-8B-Ins...
2026.05
73.2
57.4
10.2
PMET
Backbone=GPT-J-6B
2026.05
30
18
2.5
MEMIT
Backbone=GPT-J-6B
2026.05
27.2
15.5
2.3
Pre-edited
Backbone=GPT-J-6B
2026.05
0.4
0.5
14.4
Pre-edited
Backbone=Qwen3-4B-Inst...
2026.05
0.3
0.4
14
Pre-edited
Backbone=LLaMA3-8B-Ins...
2026.05
0.3
0.4
21.8
Feedback
Search any
task
Search any
task