{ "fold_results": [ { "fold": 0, "best_params": { "dropout": 0.4108948503325973, "lr": 0.0007340472227986773, "weight_decay": 0.0007067135047679004, "backbone_lr_ratio": 0.7321173243252591, "moe_n_experts": 2, "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": 14, "test_metrics": { "size": { "n_samples": 83, "mse": 1.65548933352636, "rmse": 1.2866582038468337, "mae": 0.5746945666081934, "r2": -0.013200221826040748 }, "delivery": { "n_samples": 58, "mse": 0.5966502866071148, "rmse": 0.7724314122348436, "mae": 0.600902550468414, "r2": 0.24162148150681173 }, "pdi": { "n_samples": 84, "accuracy": 0.7619047619047619, "precision": 0.7006414825374199, "recall": 0.7006414825374199, "f1": 0.7006414825374199 }, "ee": { "n_samples": 84, "accuracy": 0.7261904761904762, "precision": 0.717012617012617, "recall": 0.7409523809523809, "f1": 0.7039699318769087 }, "toxic": { "n_samples": 58, "accuracy": 0.9827586206896551, "precision": 0.9912280701754386, "recall": 0.75, "f1": 0.8289085545722714 }, "biodist": { "n_samples": 58, "kl_divergence": 0.16024890157851573, "js_divergence": 0.032467622854324446 } } }, { "fold": 1, "best_params": { "dropout": 0.24482591245144014, "lr": 0.00042458936788614755, "weight_decay": 0.00014902327669828248, "backbone_lr_ratio": 0.24808069123277332, "moe_n_experts": 2, "moe_top_k": 1, "moe_expert_hidden_mult": 1, "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": 7, "test_metrics": { "size": { "n_samples": 84, "mse": 0.3445300972529715, "rmse": 0.5869668621421242, "mae": 0.4555579432108927, "r2": -0.05207242874852769 }, "delivery": { "n_samples": 61, "mse": 1.0775295165871506, "rmse": 1.0380411921437176, "mae": 0.6340284446773471, "r2": 0.18438783389531754 }, "pdi": { "n_samples": 84, "accuracy": 0.6904761904761905, "precision": 0.5972915181753385, "recall": 0.6031746031746031, "f1": 0.5997067448680352 }, "ee": { "n_samples": 84, "accuracy": 0.5476190476190477, "precision": 0.47222222222222215, "recall": 0.4779744065458351, "f1": 0.46562819203268635 }, "toxic": { "n_samples": 61, "accuracy": 0.9508196721311475, "precision": 0.75, "recall": 0.9741379310344828, "f1": 0.8200589970501475 }, "biodist": { "n_samples": 61, "kl_divergence": 0.24783434314436234, "js_divergence": 0.0666961611545551 } } }, { "fold": 2, "best_params": { "dropout": 0.330147494187667, "lr": 0.0007391626870428583, "weight_decay": 0.08248214185502709, "backbone_lr_ratio": 0.3321485864522685, "moe_n_experts": 4, "moe_top_k": 1, "moe_expert_hidden_mult": 1, "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": 10, "test_metrics": { "size": { "n_samples": 84, "mse": 0.393067832783495, "rmse": 0.6269512204179006, "mae": 0.48754836206457447, "r2": 0.22369805318261293 }, "delivery": { "n_samples": 60, "mse": 0.6293676153394357, "rmse": 0.7933269284093637, "mae": 0.5411118950306748, "r2": 0.16220261953292037 }, "pdi": { "n_samples": 84, "accuracy": 0.6547619047619048, "precision": 0.6282346175963197, "recall": 0.6634897360703812, "f1": 0.6212097651998134 }, "ee": { "n_samples": 84, "accuracy": 0.6547619047619048, "precision": 0.6078622482131254, "recall": 0.6720354808590102, "f1": 0.6212949538882847 }, "toxic": { "n_samples": 61, "accuracy": 0.9344262295081968, "precision": 0.7142857142857143, "recall": 0.9655172413793103, "f1": 0.7821428571428571 }, "biodist": { "n_samples": 60, "kl_divergence": 0.18339582654236564, "js_divergence": 0.04379028402728257 } } }, { "fold": 3, "best_params": { "dropout": 0.2277771320023231, "lr": 0.0009696339767409517, "weight_decay": 0.0003933823864828228, "backbone_lr_ratio": 0.4073135560471746, "moe_n_experts": 4, "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": 12, "test_metrics": { "size": { "n_samples": 83, "mse": 1.732527790200732, "rmse": 1.316255214690803, "mae": 0.5408693312566324, "r2": 0.020069076700442334 }, "delivery": { "n_samples": 59, "mse": 0.5845268150974521, "rmse": 0.7645435338144271, "mae": 0.48413719012701917, "r2": 0.43177048756483927 }, "pdi": { "n_samples": 84, "accuracy": 0.7142857142857143, "precision": 0.6445623342175066, "recall": 0.6598240469208211, "f1": 0.6499999999999999 }, "ee": { "n_samples": 84, "accuracy": 0.6428571428571429, "precision": 0.5734711041691698, "recall": 0.5891402714932127, "f1": 0.5722830194879885 }, "toxic": { "n_samples": 60, "accuracy": 1.0, "precision": 1.0, "recall": 1.0, "f1": 1.0 }, "biodist": { "n_samples": 60, "kl_divergence": 0.17380496676972265, "js_divergence": 0.04216868552091467 } } }, { "fold": 4, "best_params": { "dropout": 0.3336459802315832, "lr": 0.0005171929496634874, "weight_decay": 0.00014235198757319405, "backbone_lr_ratio": 0.23125566110577128, "moe_n_experts": 4, "moe_top_k": 1, "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": 15, "test_metrics": { "size": { "n_samples": 84, "mse": 0.8359511180893293, "rmse": 0.9143036246725315, "mae": 0.5420850133467909, "r2": -0.11860643443263852 }, "delivery": { "n_samples": 58, "mse": 1.0796427545219585, "rmse": 1.03905859051449, "mae": 0.649985611872298, "r2": -0.04399672077007333 }, "pdi": { "n_samples": 84, "accuracy": 0.7261904761904762, "precision": 0.6918367346938775, "recall": 0.7412023460410557, "f1": 0.6946420104314841 }, "ee": { "n_samples": 84, "accuracy": 0.6785714285714286, "precision": 0.6383091787439613, "recall": 0.7337606837606838, "f1": 0.6521164021164022 }, "toxic": { "n_samples": 59, "accuracy": 1.0, "precision": 1.0, "recall": 1.0, "f1": 1.0 }, "biodist": { "n_samples": 58, "kl_divergence": 0.23837358595644248, "js_divergence": 0.06386943316370278 } } } ], "summary_stats": { "size": { "mse_mean": 0.9923132343705776, "mse_std": 0.5984792866289451, "rmse_mean": 0.9462270251540387, "rmse_std": 0.31139629291100435, "mae_mean": 0.5201510432974168, "mae_std": 0.04270959104369673, "r2_mean": 0.011977608975169663, "r2_std": 0.11548120300234146 }, "delivery": { "mse_mean": 0.7935433976306223, "mse_std": 0.23319925995033153, "rmse_mean": 0.8814803314233683, "rmse_std": 0.12859169080609734, "mae_mean": 0.5820331384351507, "mae_std": 0.061539044654339216, "r2_mean": 0.19519714034596314, "r2_std": 0.15267046261263886 }, "pdi": { "accuracy_mean": 0.7095238095238096, "accuracy_std": 0.03579356280564977, "precision_mean": 0.6525133374440925, "precision_std": 0.038896714345804356, "recall_mean": 0.6736664429488562, "recall_std": 0.045941327217941516, "f1_mean": 0.6532400006073505, "f1_std": 0.03965670173096385 }, "ee": { "accuracy_mean": 0.65, "accuracy_std": 0.058612064881386504, "precision_mean": 0.6017754740722191, "precision_std": 0.08029513301572835, "recall_mean": 0.6427726447222246, "recall_std": 0.09880211353738708, "f1_mean": 0.6030584998804541, "f1_std": 0.08094020366006734 }, "toxic": { "accuracy_mean": 0.9736009044657997, "accuracy_std": 0.026575657889142013, "precision_mean": 0.8911027568922305, "precision_std": 0.1303200307180046, "recall_mean": 0.9379310344827585, "recall_std": 0.09496854726841662, "f1_mean": 0.8862220817530553, "f1_std": 0.09421858471870986 }, "biodist": { "kl_divergence_mean": 0.20073152479828177, "kl_divergence_std": 0.03549650407693031, "js_divergence_mean": 0.04979843734415591, "js_divergence_std": 0.013252880014325175 } } }