hparams.json

{
  "experiment_name": "distillseq_deepseek_classes2_auto_20260618_210818",
  "model_type": "distillseq",
  "sut": "deepseek",
  "num_classes": 2,
  "device": "cuda",
  "training_config": {
    "max_epochs": 10,
    "early_stopping_patience": 50,
    "batch_size": 128,
    "num_workers": 4,
    "device": "auto",
    "gradient_clip": 1.0,
    "checkpoint_dir": "output/checkpoints",
    "log_dir": "output/logs",
    "gradient_accumulation_steps": 32,
    "use_amp": true
  },
  "model_config": {
    "model_name": "roberta-base",
    "lr": 4e-05,
    "weight_decay": 0.01,
    "warmup_steps": 0,
    "max_length": 800,
    "gradient_checkpointing": true,
    "freeze_layers": 0,
    "dropout": 0.1,
    "batch_size": 128,
    "gradient_accumulation_steps": 32,
    "use_amp": true,
    "binary_threshold": 0.7,
    "num_classes": 2,
    "loss_fn": "bce"
  }
}