mirror of
https://github.com/RYDE-WORK/lnp_ml.git
synced 2026-07-24 23:11:52 +08:00
- Updated ablation_summary.csv with full 6-variant comparison - Updated run_all.sh to run 6 variants (added BioT5 variants) - Added per-fold json artifacts (best_params/test_metrics/splits/history/summary) - Model weights (.pt) excluded per .gitignore
977 lines
28 KiB
JSON
977 lines
28 KiB
JSON
{
|
|
"fold_results": [
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.3391599915244341,
|
|
"lr": 0.0006978281265126031,
|
|
"weight_decay": 2.259279742015695e-05,
|
|
"backbone_lr_ratio": 0.024658447214487383,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 19,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.3117196588596394,
|
|
"rmse": 1.1453033043083563,
|
|
"mae": 0.5201383563184954,
|
|
"r2": 0.19719528092688665
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.5496407141982552,
|
|
"rmse": 0.7413775786994473,
|
|
"mae": 0.5149239480880827,
|
|
"r2": 0.3013734847802215
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7857142857142857,
|
|
"precision": 0.7431527693244067,
|
|
"recall": 0.7847469707769066,
|
|
"f1": 0.7543859649122806
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.6358640855011091,
|
|
"recall": 0.6842857142857143,
|
|
"f1": 0.6417137679596151
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9655172413793104,
|
|
"precision": 0.75,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8242424242424242
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.2759811566670028,
|
|
"js_divergence": 0.05473090547475484
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.1460218265559116,
|
|
"lr": 0.00016524680777017762,
|
|
"weight_decay": 3.41636179669361e-05,
|
|
"backbone_lr_ratio": 0.030281629390877935,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 28,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.3498572423658914,
|
|
"rmse": 0.5914873137827146,
|
|
"mae": 0.4678219245037153,
|
|
"r2": -0.06833963600250037
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 0.975049483099894,
|
|
"rmse": 0.9874459393302978,
|
|
"mae": 0.5932807174557057,
|
|
"r2": 0.26195783156902996
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6428571428571429,
|
|
"precision": 0.5614035087719298,
|
|
"recall": 0.5714285714285714,
|
|
"f1": 0.5625
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5119047619047619,
|
|
"precision": 0.459298871063577,
|
|
"recall": 0.4492174492174492,
|
|
"f1": 0.4371660859465738
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9508196721311475,
|
|
"precision": 0.75,
|
|
"recall": 0.9741379310344828,
|
|
"f1": 0.8200589970501475
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.4447307490351696,
|
|
"js_divergence": 0.11468280216135121
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.19165645785089888,
|
|
"lr": 0.0001935604862425119,
|
|
"weight_decay": 0.00017856185912789333,
|
|
"backbone_lr_ratio": 0.22382721014916093,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 24,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.4043717162102168,
|
|
"rmse": 0.6359022851116489,
|
|
"mae": 0.5044828815873534,
|
|
"r2": 0.2013730853809501
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.653991005253972,
|
|
"rmse": 0.8086971035276261,
|
|
"mae": 0.5367082674133902,
|
|
"r2": 0.12942461973467534
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.6165865384615384,
|
|
"recall": 0.6422287390029326,
|
|
"f1": 0.6179337231968811
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.6064814814814815,
|
|
"recall": 0.6573295985060691,
|
|
"f1": 0.6185851973086015
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.3203958948924267,
|
|
"js_divergence": 0.0790992600589471
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.2778201867797741,
|
|
"lr": 0.0004781288818954282,
|
|
"weight_decay": 0.0016987970387862138,
|
|
"backbone_lr_ratio": 0.9458696393367624,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 13,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.748584356597991,
|
|
"rmse": 1.3223404843677709,
|
|
"mae": 0.5563501338979386,
|
|
"r2": 0.010987360364530852
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.5770503693567272,
|
|
"rmse": 0.7596383148293188,
|
|
"mae": 0.4734161896315419,
|
|
"r2": 0.4390384811081155
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7023809523809523,
|
|
"precision": 0.6190306486101211,
|
|
"recall": 0.6224340175953079,
|
|
"f1": 0.6205962059620596
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6428571428571429,
|
|
"precision": 0.5750293634014564,
|
|
"recall": 0.5801659125188537,
|
|
"f1": 0.5675413711583924
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.23708880698730234,
|
|
"js_divergence": 0.058182524369121934
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.3730299609537786,
|
|
"lr": 0.0007646220605180808,
|
|
"weight_decay": 0.004588258639584098,
|
|
"backbone_lr_ratio": 0.17129190112240483,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 14,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.6858689507140188,
|
|
"rmse": 0.8281720538112951,
|
|
"mae": 0.507989236236816,
|
|
"r2": 0.08222214810857031
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 1.0067051771163806,
|
|
"rmse": 1.0033469873958762,
|
|
"mae": 0.6234078759263302,
|
|
"r2": 0.026532712520156876
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.8095238095238095,
|
|
"precision": 0.7541666666666667,
|
|
"recall": 0.7683284457478006,
|
|
"f1": 0.7605131860299359
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.613982199253517,
|
|
"recall": 0.6482905982905983,
|
|
"f1": 0.6235334654147644
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.865909090909091
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.18654112539394882,
|
|
"js_divergence": 0.045228540121851735
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.3391599915244341,
|
|
"lr": 0.0006978281265126031,
|
|
"weight_decay": 2.259279742015695e-05,
|
|
"backbone_lr_ratio": 0.024658447214487383,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 19,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.3750079031472513,
|
|
"rmse": 1.172607309864326,
|
|
"mae": 0.5439391713813845,
|
|
"r2": 0.15846131758892956
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.6388360214639206,
|
|
"rmse": 0.7992721823408597,
|
|
"mae": 0.6000871008005122,
|
|
"r2": 0.18800086685131656
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7738095238095238,
|
|
"precision": 0.7194960212201591,
|
|
"recall": 0.7359230220955096,
|
|
"f1": 0.7262905162064826
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.6170152052504995,
|
|
"recall": 0.6361904761904763,
|
|
"f1": 0.606215410813112
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9655172413793104,
|
|
"precision": 0.75,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8242424242424242
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.261216982705458,
|
|
"js_divergence": 0.05075444227253343
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.1460218265559116,
|
|
"lr": 0.00016524680777017762,
|
|
"weight_decay": 3.41636179669361e-05,
|
|
"backbone_lr_ratio": 0.030281629390877935,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 28,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.3686333012661717,
|
|
"rmse": 0.6071517942542637,
|
|
"mae": 0.48651123885065317,
|
|
"r2": -0.1256750445692565
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 0.9733903398500513,
|
|
"rmse": 0.9866054631158553,
|
|
"mae": 0.5974082425542053,
|
|
"r2": 0.26321368340329643
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6428571428571429,
|
|
"precision": 0.5830797321972003,
|
|
"recall": 0.6031746031746033,
|
|
"f1": 0.5822281167108754
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5238095238095238,
|
|
"precision": 0.4712121212121212,
|
|
"recall": 0.4748584748584748,
|
|
"f1": 0.45963079066527346
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9508196721311475,
|
|
"precision": 0.75,
|
|
"recall": 0.9741379310344828,
|
|
"f1": 0.8200589970501475
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.4177099715579412,
|
|
"js_divergence": 0.10894325162049644
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.19165645785089888,
|
|
"lr": 0.0001935604862425119,
|
|
"weight_decay": 0.00017856185912789333,
|
|
"backbone_lr_ratio": 0.22382721014916093,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 24,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.45941595200508917,
|
|
"rmse": 0.6778022956623039,
|
|
"mae": 0.536907361875776,
|
|
"r2": 0.09266170315962419
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.6760520482413604,
|
|
"rmse": 0.8222238431481785,
|
|
"mae": 0.5173251855575169,
|
|
"r2": 0.10005754781242915
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7023809523809523,
|
|
"precision": 0.6503347534996957,
|
|
"recall": 0.6810850439882699,
|
|
"f1": 0.6554552912223134
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.6228947368421053,
|
|
"recall": 0.6851073762838468,
|
|
"f1": 0.63836444060848
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.30111167063544947,
|
|
"js_divergence": 0.07488230973370555
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.2778201867797741,
|
|
"lr": 0.0004781288818954282,
|
|
"weight_decay": 0.0016987970387862138,
|
|
"backbone_lr_ratio": 0.9458696393367624,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 13,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.7221693732391328,
|
|
"rmse": 1.3123145100314684,
|
|
"mae": 0.5580621898196727,
|
|
"r2": 0.025927876284791762
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.6432654489341152,
|
|
"rmse": 0.802038308894354,
|
|
"mae": 0.48762286295009366,
|
|
"r2": 0.3746695566853039
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6904761904761905,
|
|
"precision": 0.625,
|
|
"recall": 0.6436950146627566,
|
|
"f1": 0.6298305084745763
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6190476190476191,
|
|
"precision": 0.5488385598141696,
|
|
"recall": 0.5659376571141277,
|
|
"f1": 0.5492581090407177
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.24666829388633807,
|
|
"js_divergence": 0.05944654640977567
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.3730299609537786,
|
|
"lr": 0.0007646220605180808,
|
|
"weight_decay": 0.004588258639584098,
|
|
"backbone_lr_ratio": 0.17129190112240483,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 14,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.5859281085907224,
|
|
"rmse": 0.76545941015231,
|
|
"mae": 0.4517239845701538,
|
|
"r2": 0.21595540911222322
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 1.0497243115021326,
|
|
"rmse": 1.0245605455521565,
|
|
"mae": 0.659899442809923,
|
|
"r2": -0.015066080266609161
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7261904761904762,
|
|
"precision": 0.6552542372881356,
|
|
"recall": 0.6678885630498533,
|
|
"f1": 0.6602778266221205
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6904761904761905,
|
|
"precision": 0.6406767539897621,
|
|
"recall": 0.6974358974358973,
|
|
"f1": 0.656747393535314
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.865909090909091
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.2629614988062015,
|
|
"js_divergence": 0.06764511949103554
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.3391599915244341,
|
|
"lr": 0.0006978281265126031,
|
|
"weight_decay": 2.259279742015695e-05,
|
|
"backbone_lr_ratio": 0.024658447214487383,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 19,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.3900751595506107,
|
|
"rmse": 1.1790144865736853,
|
|
"mae": 0.5758160261594387,
|
|
"r2": 0.14923978579103203
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.6073177306775106,
|
|
"rmse": 0.7793059288094186,
|
|
"mae": 0.600023343581064,
|
|
"r2": 0.22806251637797592
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7380952380952381,
|
|
"precision": 0.6987522281639929,
|
|
"recall": 0.7384176764076977,
|
|
"f1": 0.7053571428571428
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7261904761904762,
|
|
"precision": 0.6854843974409192,
|
|
"recall": 0.7452380952380953,
|
|
"f1": 0.6937398293917098
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9655172413793104,
|
|
"precision": 0.75,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8242424242424242
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.27413470235907433,
|
|
"js_divergence": 0.05513125779838101
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.1460218265559116,
|
|
"lr": 0.00016524680777017762,
|
|
"weight_decay": 3.41636179669361e-05,
|
|
"backbone_lr_ratio": 0.030281629390877935,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 28,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.36393501536520506,
|
|
"rmse": 0.603270267264354,
|
|
"mae": 0.4734607306752531,
|
|
"r2": -0.11132814977487948
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 1.0230225277101266,
|
|
"rmse": 1.0114457611311278,
|
|
"mae": 0.6079123505375913,
|
|
"r2": 0.22564569512462318
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6309523809523809,
|
|
"precision": 0.5648148148148148,
|
|
"recall": 0.5793650793650793,
|
|
"f1": 0.5635997988939165
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5357142857142857,
|
|
"precision": 0.4692511192511193,
|
|
"recall": 0.47117168545739974,
|
|
"f1": 0.45642084094096474
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.418990654299829,
|
|
"js_divergence": 0.11085459080537044
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.19165645785089888,
|
|
"lr": 0.0001935604862425119,
|
|
"weight_decay": 0.00017856185912789333,
|
|
"backbone_lr_ratio": 0.22382721014916093,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 24,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.5235245725387611,
|
|
"rmse": 0.7235499792956677,
|
|
"mae": 0.5419669135063443,
|
|
"r2": -0.03395167696776724
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.6574946725605854,
|
|
"rmse": 0.8108604519648158,
|
|
"mae": 0.5131497412997609,
|
|
"r2": 0.12476063128028725
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7142857142857143,
|
|
"precision": 0.6517857142857143,
|
|
"recall": 0.6744868035190617,
|
|
"f1": 0.6583050847457628
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.6053508771929824,
|
|
"recall": 0.6573295985060691,
|
|
"f1": 0.6168590642643937
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.32248222996451137,
|
|
"js_divergence": 0.08021451849612893
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.2778201867797741,
|
|
"lr": 0.0004781288818954282,
|
|
"weight_decay": 0.0016987970387862138,
|
|
"backbone_lr_ratio": 0.9458696393367624,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 13,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 1.8525915356133351,
|
|
"rmse": 1.3610993849140243,
|
|
"mae": 0.5908805231664733,
|
|
"r2": -0.047839892819373286
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.6323231865413851,
|
|
"rmse": 0.7951875165905116,
|
|
"mae": 0.45627199585344325,
|
|
"r2": 0.38530673579114527
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6904761904761905,
|
|
"precision": 0.616710875331565,
|
|
"recall": 0.6290322580645161,
|
|
"f1": 0.6208333333333333
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6190476190476191,
|
|
"precision": 0.5357413600891862,
|
|
"recall": 0.5468325791855203,
|
|
"f1": 0.5385907715804623
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.27107935331329597,
|
|
"js_divergence": 0.06829314930379418
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.3730299609537786,
|
|
"lr": 0.0007646220605180808,
|
|
"weight_decay": 0.004588258639584098,
|
|
"backbone_lr_ratio": 0.17129190112240483,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 14,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.6854311840545606,
|
|
"rmse": 0.8279077146968499,
|
|
"mae": 0.5337513768213934,
|
|
"r2": 0.08280793427650945
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 1.0532552367124597,
|
|
"rmse": 1.0262822402791836,
|
|
"mae": 0.6141178463425102,
|
|
"r2": -0.01848042665612204
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7976190476190477,
|
|
"precision": 0.7406779661016949,
|
|
"recall": 0.7602639296187683,
|
|
"f1": 0.7489010022859153
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6904761904761905,
|
|
"precision": 0.6378835978835978,
|
|
"recall": 0.676068376068376,
|
|
"f1": 0.6520163303668458
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.865909090909091
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.19382075032145132,
|
|
"js_divergence": 0.047516561234571965
|
|
}
|
|
}
|
|
}
|
|
],
|
|
"summary_stats": {
|
|
"size": {
|
|
"mse_mean": 0.9218076020079066,
|
|
"mse_std": 0.5523329895296802,
|
|
"rmse_mean": 0.9168921729394024,
|
|
"rmse_std": 0.28480931377040175,
|
|
"mae_mean": 0.5233201366247241,
|
|
"mae_std": 0.039438901131415294,
|
|
"r2_mean": 0.055313166724018085,
|
|
"r2_std": 0.11161236452714787
|
|
},
|
|
"delivery": {
|
|
"mse_mean": 0.7811412182145917,
|
|
"mse_std": 0.1931313144142769,
|
|
"rmse_mean": 0.8772192110406017,
|
|
"rmse_std": 0.10783169291027488,
|
|
"mae_mean": 0.5597036740534448,
|
|
"mae_std": 0.06056646385687513,
|
|
"r2_mean": 0.20096652374105642,
|
|
"r2_std": 0.1381743563430634
|
|
},
|
|
"pdi": {
|
|
"accuracy_mean": 0.7142857142857142,
|
|
"accuracy_std": 0.055498837260466956,
|
|
"precision_mean": 0.653349764982509,
|
|
"precision_std": 0.06229638797862123,
|
|
"recall_mean": 0.6734999158998424,
|
|
"recall_std": 0.06721815241332428,
|
|
"f1_mean": 0.6578005134302398,
|
|
"f1_std": 0.06528497096446977
|
|
},
|
|
"ee": {
|
|
"accuracy_mean": 0.6396825396825395,
|
|
"accuracy_std": 0.06366049088938247,
|
|
"precision_mean": 0.5816669819778402,
|
|
"precision_std": 0.0676118875849743,
|
|
"recall_mean": 0.6116972992771311,
|
|
"recall_std": 0.08879084856683034,
|
|
"f1_mean": 0.5837588579330147,
|
|
"f1_std": 0.07721106612480472
|
|
},
|
|
"toxic": {
|
|
"accuracy_mean": 0.9489467380785851,
|
|
"accuracy_std": 0.014806201481902809,
|
|
"precision_mean": 0.7433333333333333,
|
|
"precision_std": 0.03241594747641431,
|
|
"recall_mean": 0.9731959208365742,
|
|
"recall_std": 0.007847570054979899,
|
|
"f1_mean": 0.8123065675720541,
|
|
"f1_std": 0.03224418146822956
|
|
},
|
|
"biodist": {
|
|
"kl_divergence_mean": 0.29566092272169336,
|
|
"kl_divergence_std": 0.07550231859418749,
|
|
"js_divergence_mean": 0.071707051956788,
|
|
"js_divergence_std": 0.022439132651345492
|
|
}
|
|
}
|
|
} |