Benchmark Subset Selection on LAM Evaluation Benchmark 40 tasks
0.977Pearson CorrelationCombined Embedding
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Combined EmbeddingSubset Size (n)=2002026.04 | 0.977 | — | — | — | |
| Difficulty-basedSubset Size (n)=2002026.04 | 0.964 | — | — | — | |
| Combined EmbeddingSubset Size (n)=1002026.04 | 0.963 | — | — | — | |
| IRT-basedSubset Size (n)=2002026.04 | 0.96 | — | — | — | |
| Random SamplingSubset Size (n)=2002026.04 | 0.959 | — | — | — | |
| Anchor PointsSubset Size (n)=2002026.04 | 0.952 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=2002026.04 | 0.943 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=2002026.04 | 0.94 | — | — | — | |
| Anchor PointsSubset Size (n)=1002026.04 | 0.94 | — | — | — | |
| Random-Search-LearnSubset Size (n)=2002026.04 | 0.937 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=2002026.04 | 0.936 | — | — | — | |
| Combined EmbeddingSubset Size (n)=502026.04 | 0.934 | — | — | — | |
| Difficulty-basedSubset Size (n)=1002026.04 | 0.924 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=1002026.04 | 0.921 | — | — | — | |
| IRT-basedSubset Size (n)=1002026.04 | 0.919 | — | — | — | |
| Random SamplingSubset Size (n)=1002026.04 | 0.916 | — | — | — | |
| Anchor PointsSubset Size (n)=502026.04 | 0.907 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=1002026.04 | 0.904 | — | — | — | |
| Random-Search-LearnSubset Size (n)=1002026.04 | 0.894 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=1002026.04 | 0.887 | — | — | — | |
| Anchor PointsSubset Size (n)=302026.04 | 0.884 | — | — | — | |
| Combined EmbeddingSubset Size (n)=302026.04 | 0.878 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=502026.04 | 0.877 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=502026.04 | 0.87 | — | — | — | |
| IRT-basedSubset Size (n)=502026.04 | 0.864 | — | — | — | |
| Difficulty-basedSubset Size (n)=502026.04 | 0.863 | — | — | — | |
| Random SamplingSubset Size (n)=502026.04 | 0.856 | — | — | — | |
| Anchor PointsSubset Size (n)=202026.04 | 0.856 | — | — | — | |
| Combined EmbeddingSubset Size (n)=202026.04 | 0.831 | — | — | — | |
| Difficulty-basedSubset Size (n)=302026.04 | 0.811 | — | — | — | |
| Variance-basedSubset Size (n)=2002026.04 | 0.804 | — | — | — | |
| Random-Search-LearnSubset Size (n)=502026.04 | 0.803 | — | — | — | |
| Anchor PointsSubset Size (n)=102026.04 | 0.797 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=502026.04 | 0.791 | — | — | — | |
| Random SamplingSubset Size (n)=302026.04 | 0.784 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=302026.04 | 0.781 | — | — | — | |
| IRT-basedSubset Size (n)=302026.04 | 0.778 | — | — | — | |
| Difficulty-basedSubset Size (n)=202026.04 | 0.761 | — | — | — | |
| Variance-basedSubset Size (n)=1002026.04 | 0.756 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=102026.04 | 0.736 | — | — | — | |
| Random-Search-LearnSubset Size (n)=302026.04 | 0.734 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=302026.04 | 0.719 | — | — | — | |
| Random SamplingSubset Size (n)=202026.04 | 0.718 | — | — | — | |
| Variance-basedSubset Size (n)=502026.04 | 0.716 | — | — | — | |
| IRT-basedSubset Size (n)=202026.04 | 0.698 | — | — | — | |
| Random-Search-LearnSubset Size (n)=202026.04 | 0.676 | — | — | — | |
| Variance-basedSubset Size (n)=302026.04 | 0.676 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=302026.04 | 0.672 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=202026.04 | 0.656 | — | — | — | |
| Combined EmbeddingSubset Size (n)=102026.04 | 0.651 | — | — | — | |
| Variance-basedSubset Size (n)=202026.04 | 0.628 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=202026.04 | 0.627 | — | — | — | |
| Random-Search-LearnSubset Size (n)=102026.04 | 0.619 | — | — | — | |
| Difficulty-basedSubset Size (n)=102026.04 | 0.608 | — | — | — | |
| Random SamplingSubset Size (n)=102026.04 | 0.559 | — | — | — | |
| Random-Sampling-LearnSubset Size (n)=102026.04 | 0.544 | — | — | — | |
| Variance-basedSubset Size (n)=102026.04 | 0.525 | — | — | — | |
| IRT-basedSubset Size (n)=102026.04 | 0.486 | — | — | — | |
| Semantic EmbeddingSubset Size (n)=102026.04 | 0.466 | — | — | — | |
| Acoustic EmbeddingSubset Size (n)=202026.04 | 0.445 | — | — | — | |
| Acoustic Embeddingn range=[10, 200]2026.04 | — | 0.85 | 92 | 250 | |
| Anchor Pointsn range=[10, 200]2026.04 | — | 0.927 | 40 | 155 | |
| Combined Embeddingn range=[10, 200]2026.04 | — | 0.943 | 32 | 67 | |
| Difficulty-basedn range=[10, 200]2026.04 | — | 0.902 | 71 | 157 | |
| IRT-basedn range=[10, 200]2026.04 | — | 0.892 | 81 | 156 | |
| Random Samplingn range=[10, 200]2026.04 | — | 0.891 | 83 | 164 | |
| Random-Sampling-Learnn range=[10, 200]2026.04 | — | 0.854 | 119 | 300 | |
| Random-Search-Learnn range=[10, 200]2026.04 | — | 0.866 | 99 | 300 | |
| Semantic Embeddingn range=[10, 200]2026.04 | — | 0.856 | 60 | 350 | |
| Variance-basedn range=[10, 200]2026.04 | — | 0.742 | — | — |