Share your thoughts, 1 month free Claude Pro on usSee more

Code on LiveCode

83.6Accuracy

Lowest Centroid

Updated 1mo ago

Evaluation Results

Method	Links
Lowest Centroid 2026.04		83.6
Greedy Decoding 2026.04		82
Bottom Window 2026.04		81.3
Pass@1 2026.04		81.2
Self-Certainty 2026.04		80.3
Tail Confidence 2026.04		80.1
Self-Certainty 2026.04		79.5
Tail Confidence 2026.04		78.6
Lowest Centroid 2026.04		78.1
Self-Certainty 2026.04		78.1
Lowest Centroid 2026.04		78
Bottom Window 2026.04		77.9
Bottom Window 2026.04		77.7
Tail Confidence 2026.04		77.4
Lowest Centroid 2026.04		77.2
Pass@1 2026.04		77.2
Pass@1 2026.04		76.7
Self-Certainty 2026.04		76.6
Tail Confidence 2026.04		76.3
Bottom Window 2026.04		75.9
Pass@1 2026.04		74.2
Greedy Decoding 2026.04		73.2
Greedy Decoding 2026.04		72.6
Greedy Decoding 2026.04		71.6
Tail Confidence 2026.04		46.8
Bottom Window 2026.04		44.5
Lowest Centroid 2026.04		44.4
Pass@1 2026.04		40.3
Self-Certainty 2026.04		40
Greedy Decoding 2026.04		39.7