Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Recognizing Textual Entailment on RTE (Adversarial Robustness)
Loading...
100
Repair Accuracy
All-layers Full FT
-4
23
50
77
Apr 1, 2026
Repair Accuracy
Remaining Accuracy
General Accuracy
Attack Success Rate
Updated 16d ago
Evaluation Results
Method
Method
Links
Repair Accuracy
Remaining Accuracy
General Accuracy
Attack Success Rate
All-layers Full FT
Model=DistilBERT
2026.04
100
100
80.8
36.3
WARP
Model=DistilBERT
2026.04
100
100
83.8
44.1
All-layers Full FT
Model=BERT
2026.04
100
98
87.4
40.1
WARP
Model=BERT
2026.04
100
100
86.2
47.9
Same-layer LoRA
Model=DistilBERT, rank=2
2026.04
62.7
97.5
97
32.3
Same-layer LoRA
Model=BERT, rank=2
2026.04
45.5
98
99
40.6
Original
Model=BERT
2026.04
10
97.5
99
39.3
Original
Model=DistilBERT
2026.04
0
100
99.2
37.4
Feedback
Search any
task
Search any
task