training_args.json

{
  "train_file": "data/synthetic/train.jsonl",
  "validation_file": "data/synthetic/validation.jsonl",
  "test_file": null,
  "output_dir": "artifacts/tiny-router",
  "encoder_name": "microsoft/deberta-v3-small",
  "device": "auto",
  "feature_mode": "full_interaction",
  "pooling_type": "attention",
  "use_head_dependencies": true,
  "dependency_hidden_dim": 32,
  "max_length": 128,
  "recency_max": 3600,
  "batch_size": 32,
  "epochs": 20,
  "encoder_lr": 2e-05,
  "head_lr": 0.0001,
  "weight_decay": 0.01,
  "warmup_ratio": 0.1,
  "dropout": 0.1,
  "seed": 13,
  "patience": 2,
  "mixed_precision": false,
  "confidence_threshold": 0.8,
  "head_loss_weights": "{}"
}