OpenBookQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
OpenBookQA
100Attack Success Rate (ASR)
20
OpenBookQA (test)
92.31Accuracy
19
OpenBookQA
85.1Accuracy
17
OpenBookQA
47OpQA Score
15
OpenBookQA
0.157Mean Per-Step Regret
15
OpenBookQA
75.6Accuracy (Inter-layer)
15
OpenBookQA
7.2Accuracy Improvement Delta
14
OpenBookQA Official Leaderboard
95.2Accuracy
14
OpenBookQA n=991 (20%)
57.52Accuracy
13
OpenBookQA 5% n=298
53.6Accuracy
13
OpenBookQA
51.4Normalized Accuracy (OpenBookQA)
13
OpenBookQA
96Accuracy
13
OpenBookQA
98.6Score
12
OpenBookQA
95.48Accuracy (Q)
12
OpenBookQA
0.874Accuracy
12
OpenBookQA
97.58Rationale Quality (Q)
12
OpenBookQA
91.4Score
12
OpenBookQA
89.4Normalized Log Accuracy
12
OpenBookQA D^v (train)
100Accuracy
12
OpenbookQA
88Open Accuracy
12
OpenBookQA (D_eval)
75.4Accuracy
12
OpenBookQA D (train)
94.6Accuracy
12
OpenBookQA D^x (train)
93.5Accuracy
12
OpenBookQA
98Accuracy
11
OpenBookQA
1.51Empirical Generalization Error G(W)
10