{ "schema_version": 2, "benchmark": "MARBLE", "metric_scale": 1, "display_scale": 100, "models": { "MERT-v2-30s": { "repo_id": "a43992899/MERT-v2-30s", "tasks": [ { "dataset": "GTZAN", "task": "Genre classification", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 23, "hidden_state_index": 22 }, "learning_rate": 0.005, "probe_seed": 1234, "metrics": { "file_acc": 0.9172413945198059 } }, { "dataset": "GTZAN", "task": "Beat tracking", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 21, "hidden_state_index": 20 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "beat_f1": 0.9058816432952881 } }, { "dataset": "GiantSteps", "task": "Key detection", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 4, "hidden_state_index": 3 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "weighted_score": 0.6697019867549668 } }, { "dataset": "EmoMusic", "task": "Emotion recognition", "supplemental": false, "representation": { "mode": "learned_all_layer_mlp", "num_layers": 24 }, "learning_rate": 5e-05, "probe_seed": 1234, "metrics": { "valence_r2": 0.6322920322418213, "arousal_r2": 0.8000704646110535 } }, { "dataset": "Chords1217", "task": "Chord recognition", "supplemental": true, "representation": { "mode": "learned_all_layer_mlp", "num_layers": 24 }, "learning_rate": 0.0001, "probe_seed": 1234, "metrics": { "acc": 0.7847631573677063 } }, { "dataset": "MagnaTagATune", "task": "Tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 22, "hidden_state_index": 21 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.9190762042999268, "average_precision": 0.4129374325275421 } }, { "dataset": "MTG-Jamendo", "task": "Genre tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 19, "hidden_state_index": 18 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8801141381263733, "average_precision": 0.21221663057804108 } }, { "dataset": "MTG-Jamendo", "task": "Instrument tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 14, "hidden_state_index": 13 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8027445673942566, "average_precision": 0.2288619726896286 } }, { "dataset": "MTG-Jamendo", "task": "Mood/theme tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 16, "hidden_state_index": 15 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.7944093942642212, "average_precision": 0.16676680743694305 } }, { "dataset": "MTG-Jamendo", "task": "Top-50 tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 13, "hidden_state_index": 12 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8417603969573975, "average_precision": 0.3217017948627472 } } ] }, "MERT-v2-FullSong": { "repo_id": "a43992899/MERT-v2-FullSong", "tasks": [ { "dataset": "GTZAN", "task": "Genre classification", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 24, "hidden_state_index": 23 }, "learning_rate": 0.0005, "probe_seed": 1234, "metrics": { "file_acc": 0.9068965315818787 } }, { "dataset": "GTZAN", "task": "Beat tracking", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 23, "hidden_state_index": 22 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "beat_f1": 0.9057297706604004 } }, { "dataset": "GiantSteps", "task": "Key detection", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 23, "hidden_state_index": 22 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "weighted_score": 0.6705298013245033 } }, { "dataset": "EmoMusic", "task": "Emotion recognition", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 24, "hidden_state_index": 23 }, "learning_rate": 0.0005, "probe_seed": 1234, "metrics": { "valence_r2": 0.6351668834686279, "arousal_r2": 0.7814431190490723 } }, { "dataset": "Chords1217", "task": "Chord recognition", "supplemental": true, "representation": { "mode": "learned_all_layer_mlp", "num_layers": 24 }, "learning_rate": 0.0005, "probe_seed": 1234, "metrics": { "acc": 0.7779459357261658 } }, { "dataset": "MagnaTagATune", "task": "Tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 23, "hidden_state_index": 22 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.9173608422279358, "average_precision": 0.4120141565799713 } }, { "dataset": "MTG-Jamendo", "task": "Genre tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 16, "hidden_state_index": 15 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8797574043273926, "average_precision": 0.20656391978263855 } }, { "dataset": "MTG-Jamendo", "task": "Instrument tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 12, "hidden_state_index": 11 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8026924133300781, "average_precision": 0.2351299524307251 } }, { "dataset": "MTG-Jamendo", "task": "Mood/theme tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 13, "hidden_state_index": 12 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.7874246835708618, "average_precision": 0.15743640065193176 } }, { "dataset": "MTG-Jamendo", "task": "Top-50 tagging", "supplemental": false, "representation": { "mode": "fixed_layer", "layer": 22, "hidden_state_index": 21 }, "learning_rate": 0.001, "probe_seed": 1234, "metrics": { "auroc": 0.8413298726081848, "average_precision": 0.3162302076816559 } } ] } } }