Spearman rank correlation on LAM Benchmark Rankings (full) benchmark leaderboard · SOTA2 Research

Spearman rank correlation on LAM Benchmark Rankings (full)

0.769Spearman Correlation (n=10)Anchor Points

Evaluation Results

Method	Links
Anchor Points2026.04		0.769	0.831	0.858	0.879	0.897	0.926	0.896	52	—
Acoustic Embedding†2026.04		0.666	0.426	0.634	0.827	0.858	0.888	0.807	180	—
Combined Embedding‡2026.04		0.615	0.787	0.824	0.889	0.918	0.939	0.901	55	300
Random-Search-Learn2026.04		0.573	0.635	0.694	0.756	0.846	0.899	0.823	200	500
Difficulty-based2026.04		0.566	0.71	0.759	0.807	0.873	0.916	0.857	110	450
Random-Sampling-Learn2026.04		0.519	0.636	0.682	0.747	0.846	0.903	0.814	200	500
Random Sampling2026.04		0.518	0.661	0.735	0.798	0.87	0.918	0.844	145	300
Variance-based2026.04		0.497	0.591	0.638	0.678	0.718	0.766	0.707	—	—
IRT-based2026.04		0.462	0.655	0.722	0.804	0.866	0.903	0.834	176	800
Semantic Embedding2026.04		0.455	0.567	0.733	0.829	0.886	0.892	0.817	200	—