Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Abstention on SPACEREJECT 135 ID + 135 OOD (held-out seq 3, 4, 5)
Loading...
95.63
Balanced Accuracy
SEMANTIC FLIP, Q-FLIP+V-FLIP
71.4396
77.7198
84
90.2802
Jun 15, 2026
Balanced Accuracy
F1 Score
Recall
Specificity
Updated 1mo ago
Evaluation Results
Method
Method
Links
Balanced Accuracy
F1 Score
Recall
Specificity
SEMANTIC FLIP, Q-FLIP+V-FLIP
Backbone=Qwen2.5-VL-7B...
2026.06
95.63
95.59
94.67
96.59
SEMANTIC FLIP, Q-FLIP only
Backbone=Qwen2.5-VL-7B...
2026.06
95.04
94.94
93.63
96.44
C2 (Tool)
Backbone=Qwen3-8B, Pro...
2026.06
87.78
88.74
96.3
79.26
SEMANTIC FLIP, V-FLIP only
Backbone=Qwen2.5-VL-7B...
2026.06
72.37
68.67
63.85
80.89
Feedback
Search any
task
Search any
task