CSQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CSQA
82.25Task Success Rate (TSR)
6
CSQA LongGenBench (test)
86.1Accuracy (n=15)
6
CSQA random architecture
18.3ASR
6
CSQA (test)
98.2TDR
6
CSQA
83.19PIQA
6
CSQA 5 v1 (eva)
100Accuracy
6
CSQA 5 v1 (Infer)
100Accuracy
6
CSQA
69Accuracy
6
CSQA (test)
70.76F1 (Comparative)
6
CSQA Synonym Replacement WordNet-based (test)
72.1Accuracy
6
CSQA (test)
14.8Failure Rate
6
CSQA
77.62AUROC
5
CSQA LONGGENBENCH
73.5Accuracy (n=15)
5
CSQA
0.24Delta (%)
5
CSQA
78.2Preference Alignment
5
CSQA out-of-domain
49.3Accuracy
4
CSQA
43.3Adaptivity
4
CSQA (ARC-c, ARC-e, BoolQ, HellaSwag, OBQA, PIQA, SciQ, WinoG.) zero-shot CS170K
67.78Average Accuracy
4
CSQA 2
60.4Accuracy
3
CSQA
100Chain Precision
3
CSQA
0.676Score
2
CSQA
100Accuracy under Attack
2
CSQA
272LEARNED Score
2
CSQA (dev)
84.9Accuracy
2
CSQA
—Learned Rate
0