mirror of
https://github.com/RYDE-WORK/lnp_ml.git
synced 2026-07-21 21:22:05 +08:00
992 lines
28 KiB
JSON
992 lines
28 KiB
JSON
{
|
|
"fold_results": [
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.12930599750984867,
|
|
"lr": 0.0003183893957684731,
|
|
"weight_decay": 0.0011797820449629415,
|
|
"backbone_lr_ratio": 0.35598913444921026,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 8,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 20,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.27316252507783767,
|
|
"rmse": 0.5226495241343262,
|
|
"mae": 0.268634506018765,
|
|
"r2": 0.09300240292875017
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.49318943105662066,
|
|
"rmse": 0.7022744698881063,
|
|
"mae": 0.5225495728182381,
|
|
"r2": 0.37312647214479966
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.75,
|
|
"precision": 0.4107142857142857,
|
|
"recall": 0.5977751756440282,
|
|
"f1": 0.447008547008547
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7023809523809523,
|
|
"precision": 0.6658119658119658,
|
|
"recall": 0.6976190476190475,
|
|
"f1": 0.6690688363683327
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9655172413793104,
|
|
"precision": 0.75,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8242424242424242
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.20575334902695963,
|
|
"js_divergence": 0.045069148869594025
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.14959924395216218,
|
|
"lr": 0.00012401964324838103,
|
|
"weight_decay": 9.656428689903883e-05,
|
|
"backbone_lr_ratio": 0.32683834688760083,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 23,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.11199287760383045,
|
|
"rmse": 0.3346533693298641,
|
|
"mae": 0.24370252518426805,
|
|
"r2": -0.8553464014912291
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 0.9945337433308287,
|
|
"rmse": 0.9972631264269368,
|
|
"mae": 0.637184505488296,
|
|
"r2": 0.24720965117371863
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5833333333333334,
|
|
"precision": 0.2606719367588933,
|
|
"recall": 0.24206349206349206,
|
|
"f1": 0.25038520801232667
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5357142857142857,
|
|
"precision": 0.4737424924924925,
|
|
"recall": 0.47331240188383045,
|
|
"f1": 0.46359870390878144
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9508196721311475,
|
|
"precision": 0.75,
|
|
"recall": 0.9741379310344828,
|
|
"f1": 0.8200589970501475
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.42904510824573017,
|
|
"js_divergence": 0.11140172525390164
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.12302210383900847,
|
|
"lr": 0.00015510064533248203,
|
|
"weight_decay": 0.0037340196924722196,
|
|
"backbone_lr_ratio": 0.3036603338505521,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 22,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.13105160735348353,
|
|
"rmse": 0.36201050724182515,
|
|
"mae": 0.29236553396497456,
|
|
"r2": -0.4041762321007929
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.6170164560338233,
|
|
"rmse": 0.7855039503616918,
|
|
"mae": 0.518595163594,
|
|
"r2": 0.178644153319804
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5714285714285714,
|
|
"precision": 0.3343023255813954,
|
|
"recall": 0.5261136712749617,
|
|
"f1": 0.34470899470899463
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6904761904761905,
|
|
"precision": 0.6311332192269291,
|
|
"recall": 0.6916433239962653,
|
|
"f1": 0.6470105449067581
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.2793246003181208,
|
|
"js_divergence": 0.07360589998440034
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.11061601330629428,
|
|
"lr": 0.0003491022701087641,
|
|
"weight_decay": 1.2696904821623371e-05,
|
|
"backbone_lr_ratio": 0.39277239063286945,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 15,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.2974639059494916,
|
|
"rmse": 0.5454025173662949,
|
|
"mae": 0.2553033943635872,
|
|
"r2": 0.08722103632340716
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.72832486531675,
|
|
"rmse": 0.8534195130864715,
|
|
"mae": 0.5703363718970095,
|
|
"r2": 0.29198169797506535
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.3663025210084033,
|
|
"recall": 0.5624039938556068,
|
|
"f1": 0.3811802616390919
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.5903252773977808,
|
|
"recall": 0.6201608848667672,
|
|
"f1": 0.59879647539222
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.4182398191784831,
|
|
"js_divergence": 0.10764099439445415
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.41353294032230026,
|
|
"lr": 0.00018606616740103076,
|
|
"weight_decay": 9.912252717813433e-05,
|
|
"backbone_lr_ratio": 0.32820993966799533,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 25,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.2385472153185827,
|
|
"rmse": 0.4884129557235175,
|
|
"mae": 0.37417884383882793,
|
|
"r2": -0.7317567503291509
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.7980907160636062,
|
|
"rmse": 0.8933592312522473,
|
|
"mae": 0.6033944039252298,
|
|
"r2": 0.2282594525293985
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5952380952380952,
|
|
"precision": 0.33809196980096085,
|
|
"recall": 0.5263056835637481,
|
|
"f1": 0.3583862620559868
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7261904761904762,
|
|
"precision": 0.6831659493065518,
|
|
"recall": 0.7594017094017094,
|
|
"f1": 0.7065123352302344
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8659090909090909
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.4050166777635646,
|
|
"js_divergence": 0.10143992174009567
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.12930599750984867,
|
|
"lr": 0.0003183893957684731,
|
|
"weight_decay": 0.0011797820449629415,
|
|
"backbone_lr_ratio": 0.35598913444921026,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 8,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 20,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.2816563970171338,
|
|
"rmse": 0.5307131023605257,
|
|
"mae": 0.27628492734518395,
|
|
"r2": 0.06479970039267835
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.5807003895647976,
|
|
"rmse": 0.7620370001284699,
|
|
"mae": 0.564853833378132,
|
|
"r2": 0.26189476312686466
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7857142857142857,
|
|
"precision": 0.42781279178338005,
|
|
"recall": 0.5944574551131928,
|
|
"f1": 0.4533531409168081
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6547619047619048,
|
|
"precision": 0.6603426956368134,
|
|
"recall": 0.7080952380952382,
|
|
"f1": 0.6472175061309188
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9827586206896551,
|
|
"precision": 0.9912280701754386,
|
|
"recall": 0.75,
|
|
"f1": 0.8289085545722714
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.2790682434585921,
|
|
"js_divergence": 0.06757546969442413
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.14959924395216218,
|
|
"lr": 0.00012401964324838103,
|
|
"weight_decay": 9.656428689903883e-05,
|
|
"backbone_lr_ratio": 0.32683834688760083,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 23,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.08666696329506099,
|
|
"rmse": 0.2943925326754417,
|
|
"mae": 0.22798642090388707,
|
|
"r2": -0.4357809346276156
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 0.9690008398737997,
|
|
"rmse": 0.9843784027871597,
|
|
"mae": 0.626403434202075,
|
|
"r2": 0.26653621845095465
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.4880952380952381,
|
|
"precision": 0.27691029900332226,
|
|
"recall": 0.2579365079365079,
|
|
"f1": 0.24170918367346939
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5357142857142857,
|
|
"precision": 0.4703263535182041,
|
|
"recall": 0.47331240188383045,
|
|
"f1": 0.46348793258357346
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9508196721311475,
|
|
"precision": 0.75,
|
|
"recall": 0.9741379310344828,
|
|
"f1": 0.8200589970501475
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.4359101867730315,
|
|
"js_divergence": 0.11254492304386336
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.12302210383900847,
|
|
"lr": 0.00015510064533248203,
|
|
"weight_decay": 0.0037340196924722196,
|
|
"backbone_lr_ratio": 0.3036603338505521,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 22,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.13747360050361937,
|
|
"rmse": 0.3707743255723343,
|
|
"mae": 0.2868303656578064,
|
|
"r2": -0.4729858432626901
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.7104536138129312,
|
|
"rmse": 0.8428841046151785,
|
|
"mae": 0.5430462850956246,
|
|
"r2": 0.054263101423121185
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6428571428571429,
|
|
"precision": 0.45827160493827157,
|
|
"recall": 0.7127496159754224,
|
|
"f1": 0.4944194569381975
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.6060343270099368,
|
|
"recall": 0.6573295985060691,
|
|
"f1": 0.6198067632850242
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.28003378134681434,
|
|
"js_divergence": 0.0731604987811758
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.11061601330629428,
|
|
"lr": 0.0003491022701087641,
|
|
"weight_decay": 1.2696904821623371e-05,
|
|
"backbone_lr_ratio": 0.39277239063286945,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 15,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.32669708395989955,
|
|
"rmse": 0.5715742156184965,
|
|
"mae": 0.31596550309514426,
|
|
"r2": -0.0024820483050997932
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.7029380513850005,
|
|
"rmse": 0.8384140095352657,
|
|
"mae": 0.5650059677881457,
|
|
"r2": 0.31666069734708746
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.3406323185011709,
|
|
"recall": 0.5230414746543779,
|
|
"f1": 0.3490853658536585
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6547619047619048,
|
|
"precision": 0.5888888888888889,
|
|
"recall": 0.6327300150829562,
|
|
"f1": 0.5970695970695971
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.3455956570999495,
|
|
"js_divergence": 0.08991455654751294
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.41353294032230026,
|
|
"lr": 0.00018606616740103076,
|
|
"weight_decay": 9.912252717813433e-05,
|
|
"backbone_lr_ratio": 0.32820993966799533,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 25,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.2498053256210641,
|
|
"rmse": 0.49980528770818755,
|
|
"mae": 0.4012378028460911,
|
|
"r2": -0.8134860989037498
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.8713336067211406,
|
|
"rmse": 0.9334525198000917,
|
|
"mae": 0.6369200194704121,
|
|
"r2": 0.1574347863646688
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6190476190476191,
|
|
"precision": 0.32608695652173914,
|
|
"recall": 0.510752688172043,
|
|
"f1": 0.350912975912976
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6785714285714286,
|
|
"precision": 0.6324786324786325,
|
|
"recall": 0.7021367521367522,
|
|
"f1": 0.6488452445899254
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8659090909090909
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.3635802403548602,
|
|
"js_divergence": 0.09224357999051915
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 0,
|
|
"best_params": {
|
|
"dropout": 0.12930599750984867,
|
|
"lr": 0.0003183893957684731,
|
|
"weight_decay": 0.0011797820449629415,
|
|
"backbone_lr_ratio": 0.35598913444921026,
|
|
"moe_n_experts": 2,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 8,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 20,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.3048058934964392,
|
|
"rmse": 0.5520922871191366,
|
|
"mae": 0.30447895555611115,
|
|
"r2": -0.012064934220566759
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.5277170018903071,
|
|
"rmse": 0.7264413272180398,
|
|
"mae": 0.553587457165122,
|
|
"r2": 0.32923984608630563
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7261904761904762,
|
|
"precision": 0.3958333333333333,
|
|
"recall": 0.5817720530835285,
|
|
"f1": 0.43162393162393164
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.75,
|
|
"precision": 0.7124588290649997,
|
|
"recall": 0.7785714285714285,
|
|
"f1": 0.7248325686863893
|
|
},
|
|
"toxic": {
|
|
"n_samples": 58,
|
|
"accuracy": 0.9655172413793104,
|
|
"precision": 0.75,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8242424242424242
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.2720172018326682,
|
|
"js_divergence": 0.060785283409926956
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 1,
|
|
"best_params": {
|
|
"dropout": 0.14959924395216218,
|
|
"lr": 0.00012401964324838103,
|
|
"weight_decay": 9.656428689903883e-05,
|
|
"backbone_lr_ratio": 0.32683834688760083,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 23,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.10482404388180623,
|
|
"rmse": 0.32376541489449767,
|
|
"mae": 0.23482046808515275,
|
|
"r2": -0.7365828681878246
|
|
},
|
|
"delivery": {
|
|
"n_samples": 61,
|
|
"mse": 0.9723624200125729,
|
|
"rmse": 0.9860843878758921,
|
|
"mae": 0.6302215732084434,
|
|
"r2": 0.2639917446186245
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5952380952380952,
|
|
"precision": 0.28258620689655173,
|
|
"recall": 0.2698412698412698,
|
|
"f1": 0.2714159292035398
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5476190476190477,
|
|
"precision": 0.48611111111111116,
|
|
"recall": 0.4884639170353456,
|
|
"f1": 0.4753962887837127
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9508196721311475,
|
|
"precision": 0.75,
|
|
"recall": 0.9741379310344828,
|
|
"f1": 0.8200589970501475
|
|
},
|
|
"biodist": {
|
|
"n_samples": 61,
|
|
"kl_divergence": 0.4600248432068658,
|
|
"js_divergence": 0.11821523045175072
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 2,
|
|
"best_params": {
|
|
"dropout": 0.12302210383900847,
|
|
"lr": 0.00015510064533248203,
|
|
"weight_decay": 0.0037340196924722196,
|
|
"backbone_lr_ratio": 0.3036603338505521,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 1,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 22,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.12549337649644715,
|
|
"rmse": 0.35425044318454585,
|
|
"mae": 0.2802231794311887,
|
|
"r2": -0.34462155879619094
|
|
},
|
|
"delivery": {
|
|
"n_samples": 60,
|
|
"mse": 0.6592902402854742,
|
|
"rmse": 0.8119668960527111,
|
|
"mae": 0.5348013029200956,
|
|
"r2": 0.12237041942365745
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.5119047619047619,
|
|
"precision": 0.4135659661975451,
|
|
"recall": 0.6746031746031745,
|
|
"f1": 0.4122222222222222
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6547619047619048,
|
|
"precision": 0.5972649572649572,
|
|
"recall": 0.6507936507936507,
|
|
"f1": 0.6069243156199677
|
|
},
|
|
"toxic": {
|
|
"n_samples": 61,
|
|
"accuracy": 0.9344262295081968,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9655172413793103,
|
|
"f1": 0.7821428571428571
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.2563386788633516,
|
|
"js_divergence": 0.06872839042316117
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 3,
|
|
"best_params": {
|
|
"dropout": 0.11061601330629428,
|
|
"lr": 0.0003491022701087641,
|
|
"weight_decay": 1.2696904821623371e-05,
|
|
"backbone_lr_ratio": 0.39277239063286945,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 2,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 32,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 15,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 83,
|
|
"mse": 0.30582211903190704,
|
|
"rmse": 0.5530118615652896,
|
|
"mae": 0.297713055668107,
|
|
"r2": 0.06157355129088471
|
|
},
|
|
"delivery": {
|
|
"n_samples": 59,
|
|
"mse": 0.7117105624010245,
|
|
"rmse": 0.8436293987296937,
|
|
"mae": 0.5593153651077616,
|
|
"r2": 0.30813277436953124
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.36681547619047616,
|
|
"recall": 0.5662442396313364,
|
|
"f1": 0.3804035250463822
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6666666666666666,
|
|
"precision": 0.5864396721189867,
|
|
"recall": 0.6201608848667672,
|
|
"f1": 0.5949003714961162
|
|
},
|
|
"toxic": {
|
|
"n_samples": 60,
|
|
"accuracy": 0.9333333333333333,
|
|
"precision": 0.7142857142857143,
|
|
"recall": 0.9649122807017544,
|
|
"f1": 0.7818181818181817
|
|
},
|
|
"biodist": {
|
|
"n_samples": 60,
|
|
"kl_divergence": 0.31884722326826903,
|
|
"js_divergence": 0.08589330459529161
|
|
}
|
|
}
|
|
},
|
|
{
|
|
"fold": 4,
|
|
"best_params": {
|
|
"dropout": 0.41353294032230026,
|
|
"lr": 0.00018606616740103076,
|
|
"weight_decay": 9.912252717813433e-05,
|
|
"backbone_lr_ratio": 0.32820993966799533,
|
|
"moe_n_experts": 4,
|
|
"moe_top_k": 1,
|
|
"moe_expert_hidden_mult": 2,
|
|
"llm_lora_r": 16,
|
|
"d_model": 256,
|
|
"num_heads": 8,
|
|
"n_attn_layers": 4,
|
|
"fusion_strategy": "attention",
|
|
"head_hidden_dim": 128,
|
|
"set_transformer_block": "sab"
|
|
},
|
|
"epoch_mean": 25,
|
|
"test_metrics": {
|
|
"size": {
|
|
"n_samples": 84,
|
|
"mse": 0.21199504587721935,
|
|
"rmse": 0.4604291974638656,
|
|
"mae": 0.33512775670914424,
|
|
"r2": -0.5389986893953644
|
|
},
|
|
"delivery": {
|
|
"n_samples": 58,
|
|
"mse": 0.8018784776714993,
|
|
"rmse": 0.8954766762297605,
|
|
"mae": 0.6036076779509413,
|
|
"r2": 0.22459674958332065
|
|
},
|
|
"pdi": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.6190476190476191,
|
|
"precision": 0.4494096489170381,
|
|
"recall": 0.7124935995903737,
|
|
"f1": 0.47481767481767473
|
|
},
|
|
"ee": {
|
|
"n_samples": 84,
|
|
"accuracy": 0.7142857142857143,
|
|
"precision": 0.664516711833785,
|
|
"recall": 0.7427350427350428,
|
|
"f1": 0.6874551971326165
|
|
},
|
|
"toxic": {
|
|
"n_samples": 59,
|
|
"accuracy": 0.9661016949152542,
|
|
"precision": 0.8,
|
|
"recall": 0.9821428571428572,
|
|
"f1": 0.8659090909090909
|
|
},
|
|
"biodist": {
|
|
"n_samples": 58,
|
|
"kl_divergence": 0.3693123344696232,
|
|
"js_divergence": 0.09443444364888388
|
|
}
|
|
}
|
|
}
|
|
],
|
|
"summary_stats": {
|
|
"size": {
|
|
"mse_mean": 0.21249719869892153,
|
|
"mse_std": 0.083819203886272,
|
|
"rmse_mean": 0.4509291694638765,
|
|
"rmse_std": 0.09570832161071456,
|
|
"mae_mean": 0.29299021591121593,
|
|
"mae_std": 0.04711634555603193,
|
|
"r2_mean": -0.33611237791230353,
|
|
"r2_std": 0.3452040990881361
|
|
},
|
|
"delivery": {
|
|
"mse_mean": 0.7425693610280117,
|
|
"mse_std": 0.15313439940788104,
|
|
"rmse_mean": 0.8571056675991812,
|
|
"rmse_std": 0.08910238828097625,
|
|
"mae_mean": 0.5779881956006352,
|
|
"mae_std": 0.040317114677088234,
|
|
"r2_mean": 0.24162283519579483,
|
|
"r2_std": 0.08163767158765754
|
|
},
|
|
"pdi": {
|
|
"accuracy_mean": 0.634126984126984,
|
|
"accuracy_std": 0.08009190991455786,
|
|
"precision_mean": 0.36320050940978454,
|
|
"precision_std": 0.06018646749200912,
|
|
"recall_mean": 0.5239036063335376,
|
|
"recall_std": 0.14740021344728976,
|
|
"f1_mean": 0.37610884530892047,
|
|
"f1_std": 0.0760087699533074
|
|
},
|
|
"ee": {
|
|
"accuracy_mean": 0.6563492063492063,
|
|
"accuracy_std": 0.06431020501550125,
|
|
"precision_mean": 0.6032694055441358,
|
|
"precision_std": 0.07290814030935075,
|
|
"recall_mean": 0.6464310864983134,
|
|
"recall_std": 0.09580096959112427,
|
|
"f1_mean": 0.6100615120789444,
|
|
"f1_std": 0.08078656411296503
|
|
},
|
|
"toxic": {
|
|
"accuracy_mean": 0.9511890595408049,
|
|
"accuracy_std": 0.016071213375949578,
|
|
"precision_mean": 0.7617961570593149,
|
|
"precision_std": 0.0689235340767993,
|
|
"recall_mean": 0.958294443004062,
|
|
"recall_std": 0.05613748087593979,
|
|
"f1_mean": 0.8151453855878632,
|
|
"f1_std": 0.03136647604932653
|
|
},
|
|
"biodist": {
|
|
"kl_divergence_mean": 0.3412071963471256,
|
|
"kl_divergence_std": 0.07500564394419486,
|
|
"js_divergence_mean": 0.08684355805526371,
|
|
"js_divergence_std": 0.020779098028044035
|
|
}
|
|
}
|
|
} |