config_roberta.json

{
    "name": "RoBERTawithDBL",
    "n_gpu": 1,

    "arch": {
        "type": "RoBERTaDiseaseClassifier",
        "args": {
            "num_symps" : 38
        }
    },
    "data_loader": {
        "type": "MultiDiseaseDataLoader",
        "args":{
            "data_dir": "data/without_control/",
            "tokenizer_type" : "roberta-base",
            "batch_size": 64,
            "shuffle": true,
            "split" : "train",
            "bal_sample" : false,
            "control_ratio" : 0,
            "max_len" : 64,
            "uncertain" : "exclude",
            "num_workers": 2
        }
    },
    "valid_data_loader": {
        "type": "MultiDiseaseDataLoader",
        "args":{
            "data_dir": "data/without_control/",
            "tokenizer_type" : "roberta-base",
            "batch_size": 64,
            "shuffle": false,
            "split" : "val",
            "bal_sample" : false,
            "control_ratio" : 0,
            "max_len" : 64,
            "uncertain" : "exclude",
            "num_workers": 2
        }
    },
    "optimizer": {
        "type": "AdamW",
        "args":{
            "lr": 2e-5,
            "weight_decay": 1e-6
        }
    },
    "loss": {
        "name":"distirbution_balanced_loss",
        "contrastive":false,
        "contrastive_gamma":1e-5
    },
    "metrics": {
        "target" : ["auc", "accuracy", "precision", "recall", "f1_score"],
        "threshold" : 0.5
    },
    "trainer": {
        "epochs": 30,

        "save_dir": "saved/",
        "save_period": 1,
        "verbosity": 2,
        
        "monitor": "max val_macro_auc",
        "early_stop": 3,

        "tensorboard": true
    }
}