bilstm/medium/config.json

{
  "seed": 42,
  "data": {
    "train_files": [
      "/kaggle/working/data/processed/processed.txt"
    ],
    "val_files": null,
    "test_files": null,
    "batch_size": 256,
    "num_workers": 12,
    "cache_dir": ".cache/",
    "cache_format": "npz",
    "max_length": 512,
    "val_split": 0.01,
    "test_split": 0,
    "diacritic_keep_probs": [
      0.2,
      0.0,
      0.0
    ]
  },
  "modeling_config": {
    "architecture": {
      "name": "bilstm",
      "embedding_dim": 128,
      "hidden_dim": 128,
      "num_layers": 3,
      "dropout": 0.3
    },
    "loss": {
      "name": "cross_entropy",
      "label_smoothing": 0.1,
      "ignore_index": 0
    },
    "optimizer": {
      "name": "adamw",
      "lr": 0.0006,
      "weight_decay": 0.01
    },
    "scheduler": {
      "name": "cosine",
      "t_max": 50
    }
  },
  "trainer": {
    "max_epochs": 40,
    "accelerator": "auto",
    "devices": 1,
    "precision": "32",
    "log_every_n_steps": 100,
    "gradient_clip_val": 1.0,
    "loggers": [
      {
        "logger_name": "tensorboard",
        "save_dir": "lightning_logs/"
      },
      {
        "logger_name": "wandb",
        "project": "Arabic-Diacritizer",
        "log_model": "all",
        "name": "main"
      }
    ],
    "callbacks": [
      {
        "name": "model_checkpoint",
        "monitor": "val_der",
        "mode": "min",
        "save_top_k": 3,
        "filename": "diacritizer-epoch={epoch:02d}-val-der={val_der:.4f}"
      },
      {
        "name": "early_stopping",
        "monitor": "val_loss",
        "mode": "min",
        "patience": 10
      },
      {
        "name": "lr_monitor",
        "logging_interval": "step"
      }
    ]
  },
  "export": {
    "output_dir": "artifacts/",
    "dummy_input_length": 50,
    "onnx_opset_version": 15,
    "use_torch_dynamo": false
  },
  "git": {
    "branch": "main",
    "commit_hash": "7d57f90a318c335a9486f2b4680389a4978aead9"
  }
}