Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Performance prediction on MNLI source domains (out-of-domain)
Loading...
0.683
ROC AUC
Cosine distance (fine-tuned)
0.49268
0.54209
0.5915
0.64091
May 26, 2023
ROC AUC
RMSE
Updated 4d ago
Evaluation Results
Method
Method
Links
ROC AUC
RMSE
Cosine distance (fine-tuned)
Model dependency=model...
2023.05
0.683
141.9
Structural drift
Model dependency=model...
2023.05
0.531
80.6
Vocabulary, structural, semantic drift
Model dependency=model...
2023.05
0.531
81
Semantic drift
Model dependency=model...
2023.05
0.521
79.1
Combined prev. model-agnostic
Model dependency=model...
2023.05
0.514
99.8
Token frequency cross-entropy
Model dependency=model...
2023.05
0.512
96.8
Cosine distance (pre-trained)
Model dependency=model...
2023.05
0.508
107.5
Token frequency JS-div
Model dependency=model...
2023.05
0.503
118.8
Baseline (no-performance drop)
Model dependency=model...
2023.05
0.5
100
Vocabulary drift
Model dependency=model...
2023.05
0.5
81.5
Feedback
Search any
task
Search any
task