config.json

{
  "backbone": "vinai/phobert-base",
  "en_backbone": "roberta-base",
  "zero_shot_backbone": "xlm-roberta-base",
  "max_length": 256,
  "word_segment": true,
  "labels": [
    "env",
    "soc",
    "gov"
  ],
  "loss": "masked_bce",
  "pos_weight": "auto",
  "train": {
    "lr": 2.7396836004694616e-05,
    "batch_size": 16,
    "epochs": 10,
    "warmup_ratio": 0.06467163677523774,
    "weight_decay": 0.019448553647326953,
    "seeds": [
      42,
      43,
      44,
      45,
      46
    ]
  },
  "threshold_tuning": "per_label_f1",
  "tune": {
    "max_epochs": 10,
    "space": {
      "lr": [
        1e-05,
        5e-05
      ],
      "batch_size": [
        16,
        32
      ],
      "warmup_ratio": [
        0.0,
        0.2
      ],
      "weight_decay": [
        0.001,
        0.1
      ],
      "dropout": [
        0.05,
        0.3
      ]
    }
  },
  "cross_label": {
    "confidence": 0.9,
    "probs_path": "data/processed/gold/esgbert_probs.parquet",
    "table_path": "data/processed/gold/topic_labeled.parquet"
  },
  "data": {
    "masked_table": "data/processed/gold/topic_masked.parquet"
  },
  "output_dir": "outputs/models/topic",
  "heads": [
    "env",
    "soc",
    "gov"
  ],
  "dropout": 0.05249460840378278,
  "thresholds": {
    "env": 0.27,
    "soc": 0.13,
    "gov": 0.46
  }
}