{ "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": 1.566305704940777, "rmse": 1.2515213561664766, "mae": 0.6106541865583823, "r2": 0.04138235411462465 }, "delivery": { "n_samples": 58, "mse": 0.5779477061981523, "rmse": 0.7602287196614926, "mae": 0.6029246019777553, "r2": 0.2653935897935683 }, "pdi": { "n_samples": 84, "accuracy": 0.75, "precision": 0.7018518518518518, "recall": 0.733071988595866, "f1": 0.7105824446267432 }, "ee": { "n_samples": 84, "accuracy": 0.75, "precision": 0.7208073948311639, "recall": 0.7614285714285715, "f1": 0.7248345524207593 }, "toxic": { "n_samples": 58, "accuracy": 0.9655172413793104, "precision": 0.75, "recall": 0.9821428571428572, "f1": 0.8242424242424242 }, "biodist": { "n_samples": 58, "kl_divergence": 0.21251985213285735, "js_divergence": 0.04386748213277705 } } }, { "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": 0.35709959292610643, "rmse": 0.597578106130158, "mae": 0.46698603426505414, "r2": -0.09045519979354744 }, "delivery": { "n_samples": 61, "mse": 1.0490149027835551, "rmse": 1.0242142855787333, "mae": 0.6541144933062987, "r2": 0.20597134095659064 }, "pdi": { "n_samples": 84, "accuracy": 0.6785714285714286, "precision": 0.5974801061007957, "recall": 0.6111111111111112, "f1": 0.6011957095129241 }, "ee": { "n_samples": 84, "accuracy": 0.5833333333333334, "precision": 0.5340802987861811, "recall": 0.5444079729794016, "f1": 0.5238095238095238 }, "toxic": { "n_samples": 61, "accuracy": 0.9508196721311475, "precision": 0.75, "recall": 0.9741379310344828, "f1": 0.8200589970501475 }, "biodist": { "n_samples": 61, "kl_divergence": 0.35185404638391565, "js_divergence": 0.09285944999940988 } } }, { "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": 0.3574223221771879, "rmse": 0.5978480761675059, "mae": 0.44976320376043166, "r2": 0.2940972997529061 }, "delivery": { "n_samples": 60, "mse": 0.6797720885553556, "rmse": 0.8244829219306824, "mae": 0.5965622533811257, "r2": 0.0951055293826022 }, "pdi": { "n_samples": 84, "accuracy": 0.7023809523809523, "precision": 0.6503347534996957, "recall": 0.6810850439882699, "f1": 0.6554552912223134 }, "ee": { "n_samples": 84, "accuracy": 0.6785714285714286, "precision": 0.6358640855011091, "recall": 0.7037815126050421, "f1": 0.646421950315165 }, "toxic": { "n_samples": 61, "accuracy": 0.9508196721311475, "precision": 0.75, "recall": 0.9741379310344828, "f1": 0.8200589970501475 }, "biodist": { "n_samples": 60, "kl_divergence": 0.14821293243934233, "js_divergence": 0.03462949947575652 } } }, { "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": 1.6763180404940512, "rmse": 1.294727013889048, "mae": 0.5597611944561443, "r2": 0.051861739559906805 }, "delivery": { "n_samples": 59, "mse": 0.6338853934116598, "rmse": 0.7961691989845248, "mae": 0.5351308236151175, "r2": 0.3837880851060248 }, "pdi": { "n_samples": 84, "accuracy": 0.6785714285714286, "precision": 0.6072124756335282, "recall": 0.6209677419354839, "f1": 0.6110444177671068 }, "ee": { "n_samples": 84, "accuracy": 0.5833333333333334, "precision": 0.522765246449457, "recall": 0.5463298139768727, "f1": 0.5215104103741292 }, "toxic": { "n_samples": 60, "accuracy": 1.0, "precision": 1.0, "recall": 1.0, "f1": 1.0 }, "biodist": { "n_samples": 60, "kl_divergence": 0.15291185556789416, "js_divergence": 0.03702706983171318 } } }, { "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": 0.8690650468716618, "rmse": 0.932236583100911, "mae": 0.5353849688045946, "r2": -0.1629169844202072 }, "delivery": { "n_samples": 58, "mse": 1.068608919216617, "rmse": 1.0337354203163482, "mae": 0.6252993111466539, "r2": -0.033327184177486036 }, "pdi": { "n_samples": 84, "accuracy": 0.7738095238095238, "precision": 0.7163742690058479, "recall": 0.7441348973607038, "f1": 0.7262905162064826 }, "ee": { "n_samples": 84, "accuracy": 0.6666666666666666, "precision": 0.6228070175438597, "recall": 0.6837606837606837, "f1": 0.6349206349206349 }, "toxic": { "n_samples": 59, "accuracy": 0.9661016949152542, "precision": 0.8, "recall": 0.9821428571428572, "f1": 0.8659090909090909 }, "biodist": { "n_samples": 58, "kl_divergence": 0.22372876119646326, "js_divergence": 0.0564551621929834 } } } ], "summary_stats": { "size": { "mse_mean": 0.965242141481957, "mse_std": 0.5684080694545139, "rmse_mean": 0.9347822270908199, "rmse_std": 0.30236456372578385, "mae_mean": 0.5245099175689214, "mae_std": 0.059460613378625654, "r2_mean": 0.026793841842736588, "r2_std": 0.15622413268313906 }, "delivery": { "mse_mean": 0.801845802033068, "mse_std": 0.2123667454006387, "rmse_mean": 0.8877661092943562, "rmse_std": 0.11712018280992008, "mae_mean": 0.6028062966853902, "mae_std": 0.0393889285669396, "r2_mean": 0.18338627221225998, "r2_std": 0.14293661799071766 }, "pdi": { "accuracy_mean": 0.7166666666666666, "accuracy_std": 0.03868589716493314, "precision_mean": 0.6546506912183438, "precision_std": 0.0481149395489297, "recall_mean": 0.678074156598287, "recall_std": 0.05503279112301304, "f1_mean": 0.660913675867114, "f1_std": 0.050644032648896935 }, "ee": { "accuracy_mean": 0.6523809523809524, "accuracy_std": 0.06317380553057902, "precision_mean": 0.6072648086223541, "precision_std": 0.07272901245927187, "recall_mean": 0.6479417109501143, "recall_std": 0.08754948574150694, "f1_mean": 0.6102994143680424, "f1_std": 0.07796579808167317 }, "toxic": { "accuracy_mean": 0.9666516561113718, "accuracy_std": 0.017972227617721277, "precision_mean": 0.8099999999999999, "precision_std": 0.09695359714832658, "recall_mean": 0.982512315270936, "recall_std": 0.009448309250648584, "f1_mean": 0.866053901850362, "f1_std": 0.06916768085550049 }, "biodist": { "kl_divergence_mean": 0.21784548954409458, "kl_divergence_std": 0.07360209023630354, "js_divergence_mean": 0.052967732726528015, "js_divergence_std": 0.021338797362634734 } } }