{
  "by_dataset": {
    "bird-song": {
      "official_five_fold_ablation_mean": 0.7224853999999999,
      "official_five_fold_accuracy_delta": 0.03901739999999998,
      "official_five_fold_vi_pll_mean": 0.7615027999999999,
      "shortened_ablation_mean_test_accuracy": 0.678,
      "shortened_accuracy_delta": 0.03966666666666663,
      "shortened_vi_pll_mean_test_accuracy": 0.7176666666666667
    },
    "lost": {
      "official_five_fold_ablation_mean": 0.7700237999999999,
      "official_five_fold_accuracy_delta": 0.015138600000000113,
      "official_five_fold_vi_pll_mean": 0.7851624,
      "shortened_ablation_mean_test_accuracy": 0.7496296296296295,
      "shortened_accuracy_delta": 0.0503703703703704,
      "shortened_vi_pll_mean_test_accuracy": 0.7999999999999999
    }
  },
  "campaign": "official-code real-data bounded reproduction v2",
  "default_schedule_unchanged": {
    "main_loop_iterations": 1001,
    "paper_reported_epochs": [
      500,
      1000
    ],
    "warmup_loop_iterations": 501
  },
  "environment": {
    "device": "cpu",
    "numpy": "1.26.4",
    "platform": "Linux-6.18.33.2-microsoft-standard-WSL2-x86_64-with-glibc2.39",
    "python": "3.12.13",
    "scipy": "1.16.3",
    "sklearn": "1.5.1",
    "torch": "2.13.0+cpu",
    "torch_threads": 8
  },
  "instrumentation_patch_sha256": "e8f17ae7d85dfbeaea52a22ff8957ea92722b7f30d356219120dca6475c0d756",
  "max_rss_mib": 484.05078125,
  "official_commit": "dc0adebd2466cb862879ba309b30e66a37b10473",
  "official_released_csv_audit": {
    "bird-song": {
      "test_accuracy_delta_vi_minus_ablation": 0.03901739999999998,
      "vi-ablation": {
        "file": "results/bird-song_vi-ablation.csv",
        "folds": 5,
        "sha256": "504ae9ac414a72c8c45b7d3f932158fc922a74a4caa91629c7939244d078c9f3",
        "test_accuracy_mean": 0.7224853999999999,
        "test_accuracy_population_std": 0.015510104082178169,
        "test_mcc_mean": 0.6802716,
        "train_accuracy_mean": 0.7858154
      },
      "vi-pll": {
        "file": "results/bird-song_vi-pll.csv",
        "folds": 5,
        "sha256": "e560ac9cf02970bdb0b7fd9d7fcdd967aae5d7d205f87955a41e1babb54b60fe",
        "test_accuracy_mean": 0.7615027999999999,
        "test_accuracy_population_std": 0.015580385481752381,
        "test_mcc_mean": 0.725596,
        "train_accuracy_mean": 0.867097
      }
    },
    "cifar10": {
      "test_accuracy_delta_vi_minus_ablation": 0.08213999999999999,
      "vi-ablation": {
        "file": "results/cifar10_vi-ablation.csv",
        "folds": 5,
        "sha256": "a00bcb60d62c2f4dfca30f7af77847afd217c592241a38413374fb8bf698056c",
        "test_accuracy_mean": 0.8842800000000001,
        "test_accuracy_population_std": 0.0013585286158193175,
        "test_mcc_mean": 0.8820615999999999,
        "train_accuracy_mean": 0.9821626
      },
      "vi-pll": {
        "file": "results/cifar10_vi-pll.csv",
        "folds": 5,
        "sha256": "4801b6aa5b684b1ab717e7dadf7ce19a1b3edd160f29960a3ffc307320034ef0",
        "test_accuracy_mean": 0.9664200000000001,
        "test_accuracy_population_std": 0.03916526011658802,
        "test_mcc_mean": 0.9642268000000002,
        "train_accuracy_mean": 0.9914476000000001
      }
    },
    "cifar100": {
      "test_accuracy_delta_vi_minus_ablation": 0.3202399999999999,
      "vi-ablation": {
        "file": "results/cifar100_vi-ablation.csv",
        "folds": 5,
        "sha256": "2f32c9a8d3b09c5916698a5753e2957afe56dd0f2a92bd0221b326b0af2d313c",
        "test_accuracy_mean": 0.4574,
        "test_accuracy_population_std": 0.02485855989392788,
        "test_mcc_mean": 0.45415219999999995,
        "train_accuracy_mean": 0.8434963999999999
      },
      "vi-pll": {
        "file": "results/cifar100_vi-pll.csv",
        "folds": 5,
        "sha256": "470eafb37c06988e80afcf5f49e751f226082a79ee854a8c81ef0ccaf3f30cd1",
        "test_accuracy_mean": 0.7776399999999999,
        "test_accuracy_population_std": 0.005073696877031549,
        "test_mcc_mean": 0.7761578,
        "train_accuracy_mean": 0.9632153999999999
      }
    },
    "fmnist": {
      "test_accuracy_delta_vi_minus_ablation": 0.07510000000000006,
      "vi-ablation": {
        "file": "results/fmnist_vi-ablation.csv",
        "folds": 5,
        "sha256": "2b516cb473e730ea9898e6aa947d7acfcdf9584dcf97455eb97a0ca557bc2403",
        "test_accuracy_mean": 0.67364,
        "test_accuracy_population_std": 0.006577111828150716,
        "test_mcc_mean": 0.6546687999999999,
        "train_accuracy_mean": 0.9082718
      },
      "vi-pll": {
        "file": "results/fmnist_vi-pll.csv",
        "folds": 5,
        "sha256": "692d8691cd94c7f7b35227724fa4b4f1bd906842f0a6b203bed23dc62172ee89",
        "test_accuracy_mean": 0.7487400000000001,
        "test_accuracy_population_std": 0.009289262618744295,
        "test_mcc_mean": 0.7257798000000001,
        "train_accuracy_mean": 0.9207942000000001
      }
    },
    "kmnist": {
      "test_accuracy_delta_vi_minus_ablation": 0.18367999999999995,
      "vi-ablation": {
        "file": "results/kmnist_vi-ablation.csv",
        "folds": 5,
        "sha256": "0deaac66d893a09f864d357a01189819bd6aabcb5199a165933cb194e466458e",
        "test_accuracy_mean": 0.40414000000000005,
        "test_accuracy_population_std": 0.007038920371761569,
        "test_mcc_mean": 0.3671206,
        "train_accuracy_mean": 0.780351
      },
      "vi-pll": {
        "file": "results/kmnist_vi-pll.csv",
        "folds": 5,
        "sha256": "07d38755af2513e47b768bc68664f1f87aa2c70bf8572786b29a8c038a91fe5f",
        "test_accuracy_mean": 0.58782,
        "test_accuracy_population_std": 0.013390205375572095,
        "test_mcc_mean": 0.549998,
        "train_accuracy_mean": 0.9051665999999999
      }
    },
    "lost": {
      "test_accuracy_delta_vi_minus_ablation": 0.015138600000000113,
      "vi-ablation": {
        "file": "results/lost_vi-ablation.csv",
        "folds": 5,
        "sha256": "dc122e19283334f495806b65b6869413dccff9fa0727114aac71fb7b4d1d3a25",
        "test_accuracy_mean": 0.7700237999999999,
        "test_accuracy_population_std": 0.022041438305156032,
        "test_mcc_mean": 0.7412328,
        "train_accuracy_mean": 0.9017383999999999
      },
      "vi-pll": {
        "file": "results/lost_vi-pll.csv",
        "folds": 5,
        "sha256": "9b5e1f419946a5fb7a55d4c3eae0872460c6906b9fc157da8c196075573deb73",
        "test_accuracy_mean": 0.7851624,
        "test_accuracy_population_std": 0.022658168474967252,
        "test_mcc_mean": 0.7592215999999999,
        "train_accuracy_mean": 0.9251394
      }
    },
    "mir-flickr": {
      "test_accuracy_delta_vi_minus_ablation": 0.03237419999999991,
      "vi-ablation": {
        "file": "results/mir-flickr_vi-ablation.csv",
        "folds": 5,
        "sha256": "b7ab3b15bbfb73ea8b40e55da9100bda92c14669b6cfcf8c22381aced0592124",
        "test_accuracy_mean": 0.6525178,
        "test_accuracy_population_std": 0.028378408295040073,
        "test_mcc_mean": 0.5886464,
        "train_accuracy_mean": 0.6952338
      },
      "vi-pll": {
        "file": "results/mir-flickr_vi-pll.csv",
        "folds": 5,
        "sha256": "025b01750ff7ad490684671f38d9e2ea13558b455f3b9a823b6bbce5e680f558",
        "test_accuracy_mean": 0.684892,
        "test_accuracy_population_std": 0.021262655713715547,
        "test_mcc_mean": 0.6256385999999999,
        "train_accuracy_mean": 0.7312048
      }
    },
    "mnist": {
      "test_accuracy_delta_vi_minus_ablation": 0.2787599999999999,
      "vi-ablation": {
        "file": "results/mnist_vi-ablation.csv",
        "folds": 5,
        "sha256": "136646ab76148a9066fec1fbfb743299caee5fddc7dea9df9c5892ef485b6fde",
        "test_accuracy_mean": 0.5293,
        "test_accuracy_population_std": 0.01757885092945496,
        "test_mcc_mean": 0.5119868000000001,
        "train_accuracy_mean": 0.7570106000000001
      },
      "vi-pll": {
        "file": "results/mnist_vi-pll.csv",
        "folds": 5,
        "sha256": "503f25d7e78a8312f92ae5ea6ebbb7c3007d5211e7d22b8c7d80877e6f7300a0",
        "test_accuracy_mean": 0.8080599999999999,
        "test_accuracy_population_std": 0.006041390568403929,
        "test_mcc_mean": 0.7882378,
        "train_accuracy_mean": 0.848588
      }
    },
    "msrc-v2": {
      "test_accuracy_delta_vi_minus_ablation": 0.0682531999999999,
      "vi-ablation": {
        "file": "results/msrc-v2_vi-ablation.csv",
        "folds": 5,
        "sha256": "9ce2cd8c7cc75399bc8ae8d118b2bd47d9eeec74f0e441747ae6c54ec5d12a3d",
        "test_accuracy_mean": 0.5341086,
        "test_accuracy_population_std": 0.028223971319429893,
        "test_mcc_mean": 0.4919354,
        "train_accuracy_mean": 0.6240064000000001
      },
      "vi-pll": {
        "file": "results/msrc-v2_vi-pll.csv",
        "folds": 5,
        "sha256": "b180c8d06119e3a48ba9d38e4608fb1288c9df306711ff57a1a6eed9e00c5264",
        "test_accuracy_mean": 0.6023618,
        "test_accuracy_population_std": 0.024490100615554834,
        "test_mcc_mean": 0.5686742,
        "train_accuracy_mean": 0.734077
      }
    },
    "yahoo-news": {
      "test_accuracy_delta_vi_minus_ablation": 0.1457526,
      "vi-ablation": {
        "file": "results/yahoo-news_vi-ablation.csv",
        "folds": 5,
        "sha256": "af085e6ce5309d482b4e5b9eedb8d7bbbc6e1cbce5317e971b9f3856e942e6e0",
        "test_accuracy_mean": 0.49410619999999994,
        "test_accuracy_population_std": 0.005239724015632858,
        "test_mcc_mean": 0.4313508,
        "train_accuracy_mean": 0.5088189999999999
      },
      "vi-pll": {
        "file": "results/yahoo-news_vi-pll.csv",
        "folds": 5,
        "sha256": "154865abb59741ff79c12bd220ebb7c44c0cc4fc6facfaa4c836bef15b82ec76",
        "test_accuracy_mean": 0.6398588,
        "test_accuracy_population_std": 0.005750534875991952,
        "test_mcc_mean": 0.598954,
        "train_accuracy_mean": 0.8573464
      }
    }
  },
  "official_repository": "https://github.com/mathefuchs/vi-pll",
  "protocol": {
    "datasets": [
      "bird-song",
      "lost"
    ],
    "main_epochs": 120,
    "mc_samples": 5,
    "seeds": [
      42,
      43,
      44
    ],
    "train_cap": 5000,
    "warmup_epochs": 60
  },
  "runs": [
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.2708108723163605,
            "epoch": 0,
            "log_p_s_xy": -16.173489816727177,
            "log_p_x_y": 0.2934362783787712,
            "log_p_y": -2.5748081822549143,
            "log_q_y_x": -2.153297983830975,
            "loss": 26.181508894889586,
            "pxy_model": 0.1443041766361423
          },
          {
            "candidate_weight_entropy": 0.10281724482774734,
            "epoch": 119,
            "log_p_s_xy": -4.392915444989359,
            "log_p_x_y": 0.5753799662474663,
            "log_p_y": -2.368265901842425,
            "log_q_y_x": -0.6041090344229052,
            "loss": 14.207726663158786,
            "pxy_model": 0.12638849899108556
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.23563429918748635
          },
          {
            "epoch": 59,
            "rmse": 0.14410038673162456
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 1.7622679471969604,
        "mean_pairwise_decoder_distance": 0.9605048894882202
      },
      "cvae_parameters": 25702,
      "dataset": "bird-song",
      "estimated_sigma": 0.14506038849623284,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.4025229262129,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.2027934204184021e-11
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.1920928955078125e-07,
      "runtime_seconds": 75.83538559899898,
      "second_classifier_parameters": 16213,
      "seed": 42,
      "test_accuracy": 0.723,
      "test_instances": 1000,
      "test_mcc": 0.6826080665048163,
      "total_trainable_parameters": 58128,
      "train_accuracy": 0.863431715857929,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.8442761062287867,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 30.44137124092348,
          "posterior_entropy": 0.6166298389434814
        },
        {
          "epoch": 119,
          "loss": 8.01852886907516,
          "posterior_entropy": 0.19968274235725403
        }
      ],
      "candidate_constraint_max_violation": 0.9720022678375244,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "dataset": "bird-song",
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 3.5762786865234375e-07,
      "runtime_seconds": 23.84957359100008,
      "seed": 42,
      "test_accuracy": 0.678,
      "test_instances": 1000,
      "test_mcc": 0.634686927712594,
      "train_accuracy": 0.73911955977989,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.7044501198372473,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    },
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.2768377363681793,
            "epoch": 0,
            "log_p_s_xy": -16.39540793818812,
            "log_p_x_y": 0.288194453524005,
            "log_p_y": -2.5875435836853518,
            "log_q_y_x": -2.166056042717349,
            "loss": 26.467449557396673,
            "pxy_model": 0.14430939603363374
          },
          {
            "candidate_weight_entropy": 0.09859249740839005,
            "epoch": 119,
            "log_p_s_xy": -4.334124311324088,
            "log_p_x_y": 0.581076713338975,
            "log_p_y": -2.371040874911893,
            "log_q_y_x": -0.5967644866435758,
            "loss": 14.155870529913134,
            "pxy_model": 0.1262076888750626
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.22772943046603802
          },
          {
            "epoch": 59,
            "rmse": 0.1443892883496011
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 1.7702217102050781,
        "mean_pairwise_decoder_distance": 0.9625462293624878
      },
      "cvae_parameters": 25702,
      "dataset": "bird-song",
      "estimated_sigma": 0.1436564624245045,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.4025229262129,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.2027934204184021e-11
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.7881393432617188e-07,
      "runtime_seconds": 77.79380788900016,
      "second_classifier_parameters": 16213,
      "seed": 43,
      "test_accuracy": 0.719,
      "test_instances": 1000,
      "test_mcc": 0.6792958254388105,
      "total_trainable_parameters": 58128,
      "train_accuracy": 0.8699349674837419,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.8517890966322198,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 30.814381014916204,
          "posterior_entropy": 0.61581951379776
        },
        {
          "epoch": 119,
          "loss": 7.997587257815946,
          "posterior_entropy": 0.19672448933124542
        }
      ],
      "candidate_constraint_max_violation": 0.9702976942062378,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "dataset": "bird-song",
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 3.5762786865234375e-07,
      "runtime_seconds": 23.00018137799998,
      "seed": 43,
      "test_accuracy": 0.683,
      "test_instances": 1000,
      "test_mcc": 0.6388721826666496,
      "train_accuracy": 0.7478739369684843,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.7142363575176744,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    },
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.28922101855278015,
            "epoch": 0,
            "log_p_s_xy": -16.328172345315256,
            "log_p_x_y": 0.3027941380296984,
            "log_p_y": -2.581121502384063,
            "log_q_y_x": -2.1572925217690004,
            "loss": 26.340367963237146,
            "pxy_model": 0.14431963454686322
          },
          {
            "candidate_weight_entropy": 0.10308544337749481,
            "epoch": 119,
            "log_p_s_xy": -4.317468335551601,
            "log_p_x_y": 0.5742396231620542,
            "log_p_y": -2.3687866849284016,
            "log_q_y_x": -0.6072116191348722,
            "loss": 14.134589441360966,
            "pxy_model": 0.12625577046383094
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.2340078299516348
          },
          {
            "epoch": 59,
            "rmse": 0.14433669468598934
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 1.7789095640182495,
        "mean_pairwise_decoder_distance": 0.969994843006134
      },
      "cvae_parameters": 25702,
      "dataset": "bird-song",
      "estimated_sigma": 0.1464925477520259,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.4025229262129,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.2027934204184021e-11
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.7881393432617188e-07,
      "runtime_seconds": 75.5674572909993,
      "second_classifier_parameters": 16213,
      "seed": 44,
      "test_accuracy": 0.711,
      "test_instances": 1000,
      "test_mcc": 0.6707274442706338,
      "total_trainable_parameters": 58128,
      "train_accuracy": 0.8514257128564282,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.8311559461813258,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 30.660704366622433,
          "posterior_entropy": 0.610693633556366
        },
        {
          "epoch": 119,
          "loss": 7.786863526990337,
          "posterior_entropy": 0.19398967921733856
        }
      ],
      "candidate_constraint_max_violation": 0.970436155796051,
      "classes_after_official_filter": 13,
      "classifier_parameters": 16213,
      "dataset": "bird-song",
      "features_after_official_filter": 38,
      "full_instances": 4998,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.17158579826355,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 4.172325134277344e-07,
      "runtime_seconds": 23.32986260400139,
      "seed": 44,
      "test_accuracy": 0.673,
      "test_instances": 1000,
      "test_mcc": 0.6280691416621966,
      "train_accuracy": 0.7368684342171086,
      "train_cap": 5000,
      "train_instances": 3998,
      "train_mcc": 0.7010679694817208,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    },
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.15046530961990356,
            "epoch": 0,
            "log_p_s_xy": -19.270621708461217,
            "log_p_x_y": 0.6419406448091779,
            "log_p_y": -2.6742364849363054,
            "log_q_y_x": -2.3874726125172208,
            "loss": -15.19903836931501,
            "pxy_model": 0.12173546624446008
          },
          {
            "candidate_weight_entropy": 0.14434659481048584,
            "epoch": 119,
            "log_p_s_xy": -3.900912914957319,
            "log_p_x_y": 0.6794598272868565,
            "log_p_y": -2.5575463260923113,
            "log_q_y_x": -0.8986693024635315,
            "loss": -30.95614733014788,
            "pxy_model": 0.12076304395414368
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.13617287380058787
          },
          {
            "epoch": 59,
            "rmse": 0.12180864659672788
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 0.6741830706596375,
        "mean_pairwise_decoder_distance": 0.5169421434402466
      },
      "cvae_parameters": 43948,
      "dataset": "lost",
      "estimated_sigma": 0.12118840773035228,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.608857796452674,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.1102230246251565e-16
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.1920928955078125e-07,
      "runtime_seconds": 21.29557384099826,
      "second_classifier_parameters": 37514,
      "seed": 42,
      "test_accuracy": 0.7866666666666666,
      "test_instances": 225,
      "test_mcc": 0.7587857130444665,
      "total_trainable_parameters": 118976,
      "train_accuracy": 0.9096989966555183,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8984177459761418,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 36.31639562334333,
          "posterior_entropy": 0.7571771144866943
        },
        {
          "epoch": 119,
          "loss": 6.836584057126727,
          "posterior_entropy": 0.3071111738681793
        }
      ],
      "candidate_constraint_max_violation": 0.8402968049049377,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "dataset": "lost",
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 3.5762786865234375e-07,
      "runtime_seconds": 5.929461564999656,
      "seed": 42,
      "test_accuracy": 0.7422222222222222,
      "test_instances": 225,
      "test_mcc": 0.7082425084809693,
      "train_accuracy": 0.8617614269788183,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8438786667189287,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    },
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.16198872029781342,
            "epoch": 0,
            "log_p_s_xy": -19.22972297668457,
            "log_p_x_y": 0.6386122192655291,
            "log_p_y": -2.673332231385367,
            "log_q_y_x": -2.385913065501622,
            "loss": -15.071918215070452,
            "pxy_model": 0.12183219179456341
          },
          {
            "candidate_weight_entropy": 0.14957979321479797,
            "epoch": 119,
            "log_p_s_xy": -4.004478795187814,
            "log_p_x_y": 0.6767580807209015,
            "log_p_y": -2.5565474033355713,
            "log_q_y_x": -0.9191818194729942,
            "loss": -30.73798656463623,
            "pxy_model": 0.12069734300549294
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.1360717764413006
          },
          {
            "epoch": 59,
            "rmse": 0.12189789056578378
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 0.7016123533248901,
        "mean_pairwise_decoder_distance": 0.5112101435661316
      },
      "cvae_parameters": 43948,
      "dataset": "lost",
      "estimated_sigma": 0.11756013757205908,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.608857796452674,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.1102230246251565e-16
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.1920928955078125e-07,
      "runtime_seconds": 23.520829045999562,
      "second_classifier_parameters": 37514,
      "seed": 43,
      "test_accuracy": 0.8044444444444444,
      "test_instances": 225,
      "test_mcc": 0.7793682729730292,
      "total_trainable_parameters": 118976,
      "train_accuracy": 0.903010033444816,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8906299650249241,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 36.30288124084473,
          "posterior_entropy": 0.7567180395126343
        },
        {
          "epoch": 119,
          "loss": 6.661709308624268,
          "posterior_entropy": 0.30992892384529114
        }
      ],
      "candidate_constraint_max_violation": 0.8136563897132874,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "dataset": "lost",
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 4.76837158203125e-07,
      "runtime_seconds": 6.960834924000665,
      "seed": 43,
      "test_accuracy": 0.7288888888888889,
      "test_instances": 225,
      "test_mcc": 0.6958009330671128,
      "train_accuracy": 0.855072463768116,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8367646264212416,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    },
    {
      "algorithm": "vi-pll",
      "audit_trace": {
        "main": [
          {
            "candidate_weight_entropy": 0.16210441291332245,
            "epoch": 0,
            "log_p_s_xy": -19.23987020765032,
            "log_p_x_y": 0.6429869490010398,
            "log_p_y": -2.672427569116865,
            "log_q_y_x": -2.38821804523468,
            "loss": -15.282821655273438,
            "pxy_model": 0.1218777564954407
          },
          {
            "candidate_weight_entropy": 0.1467060148715973,
            "epoch": 119,
            "log_p_s_xy": -4.055566651480539,
            "log_p_x_y": 0.6794576006276267,
            "log_p_y": -2.556765386036464,
            "log_q_y_x": -0.9188610655920846,
            "loss": -30.82157584599086,
            "pxy_model": 0.12074240149111233
          }
        ],
        "warmup": [
          {
            "epoch": 0,
            "rmse": 0.13782092395532877
          },
          {
            "epoch": 59,
            "rmse": 0.12190104501576896
          }
        ]
      },
      "candidate_constraint_max_violation": 0.0,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "cvae_label_conditioning_probe": {
        "max_pairwise_decoder_distance": 0.7064527273178101,
        "mean_pairwise_decoder_distance": 0.5155908465385437
      },
      "cvae_parameters": 43948,
      "dataset": "lost",
      "estimated_sigma": 0.12315943971947536,
      "factorization_terms_executed": [
        "log Q(Y|X)",
        "log P(Y)",
        "log P(X|Y)",
        "log P(S|X,Y)"
      ],
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "max_entropy_prior": {
        "entropy": 2.608857796452674,
        "max_bound_violation": 0.0,
        "min_bound_violation": 0.0,
        "sum_residual": 1.1102230246251565e-16
      },
      "mc_samples": 5,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 1.1920928955078125e-07,
      "runtime_seconds": 26.15351639599976,
      "second_classifier_parameters": 37514,
      "seed": 44,
      "test_accuracy": 0.8088888888888889,
      "test_instances": 225,
      "test_mcc": 0.784477934804284,
      "total_trainable_parameters": 118976,
      "train_accuracy": 0.9085841694537347,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8970442211588884,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 60
    },
    {
      "algorithm": "vi-ablation",
      "audit_trace": [
        {
          "epoch": 0,
          "loss": 36.23127365112305,
          "posterior_entropy": 0.7565066814422607
        },
        {
          "epoch": 119,
          "loss": 7.287430627005441,
          "posterior_entropy": 0.30685684084892273
        }
      ],
      "candidate_constraint_max_violation": 0.8275954723358154,
      "classes_after_official_filter": 14,
      "classifier_parameters": 37514,
      "dataset": "lost",
      "features_after_official_filter": 108,
      "full_instances": 1122,
      "main_epochs_executed": 120,
      "mc_samples": 10,
      "mean_candidate_set_size": 2.218506097793579,
      "official_fold": 0,
      "posterior_rowsum_max_residual": 4.172325134277344e-07,
      "runtime_seconds": 7.156720177998068,
      "seed": 44,
      "test_accuracy": 0.7777777777777778,
      "test_instances": 225,
      "test_mcc": 0.7499112288128404,
      "train_accuracy": 0.8494983277591973,
      "train_cap": 5000,
      "train_instances": 897,
      "train_mcc": 0.8304056931793165,
      "true_label_in_candidate_rate": 1.0,
      "warmup_epochs_executed": 0
    }
  ],
  "runtime_seconds": 390.83336939700166
}