{
  "created_at": "2026-05-19T20:13:44.207534+00:00",
  "config": {
    "hf_dataset": "faodl/amis-agri-soybeans",
    "hf_subset": null,
    "train_split": "train",
    "validation_split": "validation",
    "test_split": "test",
    "text_col": "chunk_text",
    "label_col": "label",
    "group_col": "id",
    "id_col": "chunk_id",
    "model_name": "distilbert/distilbert-base-multilingual-cased",
    "output_dir": "/content/agri-soybeans-classifier",
    "max_length": 256,
    "learning_rate": 2e-05,
    "weight_decay": 0.01,
    "num_train_epochs": 5.0,
    "per_device_train_batch_size": 16,
    "per_device_eval_batch_size": 32,
    "gradient_accumulation_steps": 1,
    "warmup_ratio": 0.1,
    "early_stopping_patience": 2,
    "seed": 42,
    "metric_for_best_model": "f1",
    "skip_transformer": false,
    "skip_baselines": false,
    "baseline_models": [
      "logistic",
      "xgboost",
      "embedding-logistic",
      "embedding-svm",
      "embedding-lightgbm"
    ],
    "tfidf_max_features": 50000,
    "embedding_model_name": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
    "embedding_batch_size": 64,
    "positive_label_name": "RELEVANT",
    "negative_label_name": "NOT_RELEVANT",
    "push_to_hub": true,
    "hub_model_id": "faodl/agri-soybeans-classifier",
    "hub_private_repo": false
  },
  "dataset_summary": {
    "train": {
      "rows": 4745,
      "labels": {
        "0": 3860,
        "1": 885
      },
      "unique_groups": 2244,
      "text_length_mean": 702.3989462592202,
      "text_length_median": 794.0
    },
    "validation": {
      "rows": 1034,
      "labels": {
        "0": 782,
        "1": 252
      },
      "unique_groups": 481,
      "text_length_mean": 710.2852998065764,
      "text_length_median": 795.0
    },
    "test": {
      "rows": 1074,
      "labels": {
        "0": 889,
        "1": 185
      },
      "unique_groups": 482,
      "text_length_mean": 708.6173184357542,
      "text_length_median": 794.0
    }
  },
  "results": [
    {
      "model_type": "logistic_tfidf",
      "model_name": "logistic",
      "artifact_dir": "/content/agri-soybeans-classifier/baselines/logistic",
      "artifact_file": "/content/agri-soybeans-classifier/baselines/logistic/logistic_tfidf.joblib",
      "validation_best_threshold": {
        "threshold": 0.45412653657308116,
        "f1": 0.8698884758364314,
        "precision": 0.8181818181818182,
        "recall": 0.9285714285714286
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.9441340782122905,
        "precision": 0.8048780487804879,
        "recall": 0.8918918918918919,
        "f1": 0.8461538461538461,
        "confusion_matrix": [
          [
            849,
            40
          ],
          [
            20,
            165
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9769850402761795,
            "recall": 0.9550056242969629,
            "f1-score": 0.9658703071672355,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8048780487804879,
            "recall": 0.8918918918918919,
            "f1-score": 0.8461538461538461,
            "support": 185.0
          },
          "accuracy": 0.9441340782122905,
          "macro avg": {
            "precision": 0.8909315445283337,
            "recall": 0.9234487580944274,
            "f1-score": 0.9060120766605408,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9473390501209625,
            "recall": 0.9441340782122905,
            "f1-score": 0.9452487566202364,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9665278326695649,
        "average_precision": 0.9143444807696574
      },
      "test_optimal_threshold": {
        "threshold": 0.45412653657308116,
        "accuracy": 0.9413407821229051,
        "precision": 0.7850467289719626,
        "recall": 0.9081081081081082,
        "f1": 0.8421052631578947,
        "confusion_matrix": [
          [
            843,
            46
          ],
          [
            17,
            168
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9802325581395349,
            "recall": 0.9482564679415073,
            "f1-score": 0.9639794168096055,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.7850467289719626,
            "recall": 0.9081081081081082,
            "f1-score": 0.8421052631578947,
            "support": 185.0
          },
          "accuracy": 0.9413407821229051,
          "macro avg": {
            "precision": 0.8826396435557488,
            "recall": 0.9281822880248077,
            "f1-score": 0.9030423399837502,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9466111629849717,
            "recall": 0.9413407821229051,
            "f1-score": 0.9429861966740687,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9665278326695649,
        "average_precision": 0.9143444807696574
      }
    },
    {
      "model_type": "xgboost_tfidf",
      "model_name": "xgboost",
      "artifact_dir": "/content/agri-soybeans-classifier/baselines/xgboost",
      "artifact_file": "/content/agri-soybeans-classifier/baselines/xgboost/xgboost_tfidf.joblib",
      "validation_best_threshold": {
        "threshold": 0.5492749810218811,
        "f1": 0.9000000000000001,
        "precision": 0.907258064516129,
        "recall": 0.8928571428571429
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.9543761638733705,
        "precision": 0.8953488372093024,
        "recall": 0.8324324324324325,
        "f1": 0.8627450980392157,
        "confusion_matrix": [
          [
            871,
            18
          ],
          [
            31,
            154
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9656319290465631,
            "recall": 0.9797525309336333,
            "f1-score": 0.972640982691234,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8953488372093024,
            "recall": 0.8324324324324325,
            "f1-score": 0.8627450980392157,
            "support": 185.0
          },
          "accuracy": 0.9543761638733705,
          "macro avg": {
            "precision": 0.9304903831279328,
            "recall": 0.9060924816830329,
            "f1-score": 0.9176930403652248,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9535254374358618,
            "recall": 0.9543761638733705,
            "f1-score": 0.9537110584262215,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9635028729516919,
        "average_precision": 0.8960212274222826
      },
      "test_optimal_threshold": {
        "threshold": 0.5492749810218811,
        "accuracy": 0.9553072625698324,
        "precision": 0.9053254437869822,
        "recall": 0.827027027027027,
        "f1": 0.864406779661017,
        "confusion_matrix": [
          [
            873,
            16
          ],
          [
            32,
            153
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9646408839779006,
            "recall": 0.9820022497187851,
            "f1-score": 0.9732441471571907,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.9053254437869822,
            "recall": 0.827027027027027,
            "f1-score": 0.864406779661017,
            "support": 185.0
          },
          "accuracy": 0.9553072625698324,
          "macro avg": {
            "precision": 0.9349831638824414,
            "recall": 0.9045146383729061,
            "f1-score": 0.9188254634091038,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9544236061051633,
            "recall": 0.9553072625698324,
            "f1-score": 0.9544965559218163,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9635028729516919,
        "average_precision": 0.8960212274222826
      }
    },
    {
      "model_type": "embedding-logistic_sentence_embeddings",
      "model_name": "logistic",
      "embedding_model_name": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
      "artifact_dir": "/content/agri-soybeans-classifier/baselines/embedding-logistic",
      "artifact_file": "/content/agri-soybeans-classifier/baselines/embedding-logistic/embedding-logistic.joblib",
      "validation_best_threshold": {
        "threshold": 0.6469514995226564,
        "f1": 0.8506616257088847,
        "precision": 0.8122743682310469,
        "recall": 0.8928571428571429
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.9385474860335196,
        "precision": 0.7531914893617021,
        "recall": 0.9567567567567568,
        "f1": 0.8428571428571429,
        "confusion_matrix": [
          [
            831,
            58
          ],
          [
            8,
            177
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9904648390941597,
            "recall": 0.9347581552305961,
            "f1-score": 0.9618055555555556,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.7531914893617021,
            "recall": 0.9567567567567568,
            "f1-score": 0.8428571428571429,
            "support": 185.0
          },
          "accuracy": 0.9385474860335196,
          "macro avg": {
            "precision": 0.8718281642279309,
            "recall": 0.9457574559936764,
            "f1-score": 0.9023313492063492,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9495937313655706,
            "recall": 0.9385474860335196,
            "f1-score": 0.941316303833762,
            "support": 1074.0
          }
        },
        "roc_auc": 0.98785151856018,
        "average_precision": 0.951385005586427
      },
      "test_optimal_threshold": {
        "threshold": 0.6469514995226564,
        "accuracy": 0.9543761638733705,
        "precision": 0.8366336633663366,
        "recall": 0.9135135135135135,
        "f1": 0.8733850129198967,
        "confusion_matrix": [
          [
            856,
            33
          ],
          [
            16,
            169
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.981651376146789,
            "recall": 0.9628796400449944,
            "f1-score": 0.9721749006246451,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8366336633663366,
            "recall": 0.9135135135135135,
            "f1-score": 0.8733850129198967,
            "support": 185.0
          },
          "accuracy": 0.9543761638733705,
          "macro avg": {
            "precision": 0.9091425197565628,
            "recall": 0.9381965767792539,
            "f1-score": 0.9227799567722709,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9566716025300445,
            "recall": 0.9543761638733705,
            "f1-score": 0.9551580205265273,
            "support": 1074.0
          }
        },
        "roc_auc": 0.98785151856018,
        "average_precision": 0.951385005586427
      }
    },
    {
      "model_type": "embedding-svm_sentence_embeddings",
      "model_name": "svm",
      "embedding_model_name": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
      "artifact_dir": "/content/agri-soybeans-classifier/baselines/embedding-svm",
      "artifact_file": "/content/agri-soybeans-classifier/baselines/embedding-svm/embedding-svm.joblib",
      "validation_best_threshold": {
        "threshold": 0.37877199430477043,
        "f1": 0.8403041825095057,
        "precision": 0.8065693430656934,
        "recall": 0.876984126984127
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.957169459962756,
        "precision": 0.8839779005524862,
        "recall": 0.8648648648648649,
        "f1": 0.8743169398907104,
        "confusion_matrix": [
          [
            868,
            21
          ],
          [
            25,
            160
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9720044792833147,
            "recall": 0.9763779527559056,
            "f1-score": 0.9741863075196409,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8839779005524862,
            "recall": 0.8648648648648649,
            "f1-score": 0.8743169398907104,
            "support": 185.0
          },
          "accuracy": 0.957169459962756,
          "macro avg": {
            "precision": 0.9279911899179004,
            "recall": 0.9206214088103852,
            "f1-score": 0.9242516237051757,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.956841614231915,
            "recall": 0.957169459962756,
            "f1-score": 0.9569834834867246,
            "support": 1074.0
          }
        },
        "roc_auc": 0.98772383181832,
        "average_precision": 0.9492625541246242
      },
      "test_optimal_threshold": {
        "threshold": 0.37877199430477043,
        "accuracy": 0.9553072625698324,
        "precision": 0.8477157360406091,
        "recall": 0.9027027027027027,
        "f1": 0.8743455497382199,
        "confusion_matrix": [
          [
            859,
            30
          ],
          [
            18,
            167
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9794754846066135,
            "recall": 0.9662542182227222,
            "f1-score": 0.9728199320498301,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8477157360406091,
            "recall": 0.9027027027027027,
            "f1-score": 0.8743455497382199,
            "support": 185.0
          },
          "accuracy": 0.9553072625698324,
          "macro avg": {
            "precision": 0.9135956103236114,
            "recall": 0.9344784604627124,
            "f1-score": 0.9235827408940249,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9567794385314637,
            "recall": 0.9553072625698324,
            "f1-score": 0.9558573987838637,
            "support": 1074.0
          }
        },
        "roc_auc": 0.98772383181832,
        "average_precision": 0.9492625541246242
      }
    },
    {
      "model_type": "embedding-lightgbm_sentence_embeddings",
      "model_name": "lightgbm",
      "embedding_model_name": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
      "artifact_dir": "/content/agri-soybeans-classifier/baselines/embedding-lightgbm",
      "artifact_file": "/content/agri-soybeans-classifier/baselines/embedding-lightgbm/embedding-lightgbm.joblib",
      "validation_best_threshold": {
        "threshold": 0.42851684417216673,
        "f1": 0.8473895582329317,
        "precision": 0.8577235772357723,
        "recall": 0.8373015873015873
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.9590316573556797,
        "precision": 0.8938547486033519,
        "recall": 0.8648648648648649,
        "f1": 0.8791208791208791,
        "confusion_matrix": [
          [
            870,
            19
          ],
          [
            25,
            160
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9720670391061452,
            "recall": 0.9786276715410573,
            "f1-score": 0.9753363228699552,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8938547486033519,
            "recall": 0.8648648648648649,
            "f1-score": 0.8791208791208791,
            "support": 185.0
          },
          "accuracy": 0.9590316573556797,
          "macro avg": {
            "precision": 0.9329608938547486,
            "recall": 0.9217462682029611,
            "f1-score": 0.9272286009954172,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9585947171852729,
            "recall": 0.9590316573556797,
            "f1-score": 0.9587628991329169,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9852369805125711,
        "average_precision": 0.9499432525904906
      },
      "test_optimal_threshold": {
        "threshold": 0.42851684417216673,
        "accuracy": 0.9590316573556797,
        "precision": 0.8895027624309392,
        "recall": 0.8702702702702703,
        "f1": 0.8797814207650273,
        "confusion_matrix": [
          [
            869,
            20
          ],
          [
            24,
            161
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.973124300111982,
            "recall": 0.9775028121484814,
            "f1-score": 0.9753086419753086,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8895027624309392,
            "recall": 0.8702702702702703,
            "f1-score": 0.8797814207650273,
            "support": 185.0
          },
          "accuracy": 0.9590316573556797,
          "macro avg": {
            "precision": 0.9313135312714607,
            "recall": 0.9238865412093759,
            "f1-score": 0.927545031370168,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9587202177367556,
            "recall": 0.9590316573556797,
            "f1-score": 0.9588537668133889,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9852369805125711,
        "average_precision": 0.9499432525904906
      }
    },
    {
      "model_type": "transformer",
      "model_name": "distilbert/distilbert-base-multilingual-cased",
      "artifact_dir": "/content/agri-soybeans-classifier/transformer",
      "validation_best_threshold": {
        "threshold": 0.4926762878894806,
        "f1": 0.9236790606653621,
        "precision": 0.9111969111969112,
        "recall": 0.9365079365079365
      },
      "test_default_0_5": {
        "threshold": 0.5,
        "accuracy": 0.9543761638733705,
        "precision": 0.8820224719101124,
        "recall": 0.8486486486486486,
        "f1": 0.8650137741046832,
        "confusion_matrix": [
          [
            868,
            21
          ],
          [
            28,
            157
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.96875,
            "recall": 0.9763779527559056,
            "f1-score": 0.9725490196078431,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.8820224719101124,
            "recall": 0.8486486486486486,
            "f1-score": 0.8650137741046832,
            "support": 185.0
          },
          "accuracy": 0.9543761638733705,
          "macro avg": {
            "precision": 0.9253862359550562,
            "recall": 0.9125133007022771,
            "f1-score": 0.9187813968562631,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9538109006549076,
            "recall": 0.9543761638733705,
            "f1-score": 0.9540257231291798,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9756270331073482,
        "average_precision": 0.9289986926848062
      },
      "test_optimal_threshold": {
        "threshold": 0.4926762878894806,
        "accuracy": 0.9553072625698324,
        "precision": 0.88268156424581,
        "recall": 0.8540540540540541,
        "f1": 0.8681318681318682,
        "confusion_matrix": [
          [
            868,
            21
          ],
          [
            27,
            158
          ]
        ],
        "classification_report": {
          "NOT_RELEVANT": {
            "precision": 0.9698324022346368,
            "recall": 0.9763779527559056,
            "f1-score": 0.9730941704035875,
            "support": 889.0
          },
          "RELEVANT": {
            "precision": 0.88268156424581,
            "recall": 0.8540540540540541,
            "f1-score": 0.8681318681318682,
            "support": 185.0
          },
          "accuracy": 0.9553072625698324,
          "macro avg": {
            "precision": 0.9262569832402234,
            "recall": 0.9152160034049799,
            "f1-score": 0.9206130192677278,
            "support": 1074.0
          },
          "weighted avg": {
            "precision": 0.9548203863799507,
            "recall": 0.9553072625698324,
            "f1-score": 0.9550140717813639,
            "support": 1074.0
          }
        },
        "roc_auc": 0.9756270331073482,
        "average_precision": 0.9289986926848062
      }
    }
  ]
}