Zero-shot Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
68.3Avg Accuracy
24
Jul 3, 2026
69.19ARC-e
23
May 7, 2026
67.3Average Performance
22
Feb 26, 2026
70.9Average Performance
21
Feb 26, 2026
71.1BoolQ Accuracy (Zero-shot)
20
Mar 6, 2026
69.4SciQ Accuracy
19
Jun 2, 2026
59.52Accuracy
18
Jun 4, 2026
71.75Accuracy
17
May 13, 2026
48CORE_V2 Score
17
Feb 26, 2026
75.97Average Score
16
May 27, 2026
68.36Mean Zero-Shot Accuracy
16
May 8, 2026
81Avg Zero-shot Accuracy
16
Apr 10, 2026
35.96ScienceQA
15
May 27, 2026
81.6PI Score (Zero-shot)
13
Jun 16, 2026
49Average Accuracy
12
Jun 2, 2026
17.8LAMBADA Perplexity (PPL)
10
May 20, 2026
0.268CORE Score
9
Jun 26, 2026
9 Zero-shot Tasks (BoolQ, HellaSwag, LAMBADA, OBQA, PIQA, SIQA, WinoGrande, ARC-Easy, ARC-Challenge)
0.7110-shot Avg Accuracy
9
May 27, 2026
76.07Winogrande Accuracy
9
Apr 10, 2026
62.47PiQA Accuracy
9
Mar 25, 2026
28.99PTB Perplexity
8
Mar 12, 2026
63.47Mean Zero-shot Accuracy
8
Feb 26, 2026
73Average Accuracy (Zero-shot)
7
Jul 7, 2026
70.7Reasoning Accuracy
7
Feb 26, 2026
55.6ARC-C Accuracy
6
Apr 21, 2026