Macro Average
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Macro-average (Mathematics, Multi-Hop QA, Code Generation)
69Accuracy
21
Macro Average (Across Datasets)
96.9AUC
18
Macro Average AIME2024, MATH, Minerva, Olympiad-Bench
55Pass@1
16
Macro Average Selected Benchmarks
52.8Pass@1 (Avg@32)
14
Macro-Average
82.14Mean Accuracy
11
Macro-average SICKR-STS, STS-B, WMT_RU_EN, WMT_EN_ZH, WMT_SI_EN (test)
76.3Pearson Correlation (r)
11
Macro-average
36.6Avg@8
10
Macro-average (MMLU, MATH, GSM8K, BBH)
46Cost Reduction
8
Macro-average 5 NDAs (Avg)
93.4Accuracy
6
Macro-average over six benchmarks
40.7C2%
6
Macro Average (PubMedQA, WebShop, ToolEmu)
98.4Utilization (Util.)
5
Macro Average Zero-shot
69.7Macro Accuracy (Zero-shot)
4
Macro Average In-domain
56.3Macro Accuracy
4