{
  "model": {
    "vocab_size": 50257,
    "d_model": 512,
    "num_layers": 6,
    "num_heads": 8,
    "d_ff": 2048,
    "max_seq_len": 512,
    "dropout": 0.1,
    "activation": "gelu",
    "layer_norm_eps": 1e-5,
    "bias": false,
    "tie_weights": true,
    "use_optimized_attention": true
  },
  "training": {
    "batch_size": 16,
    "max_epochs": 5,
    "learning_rate": 2e-4,
    "weight_decay": 0.01,
    "warmup_steps": 500,
    "max_grad_norm": 1.0,
    "gradient_accumulation_steps": 8,
    "use_amp": true,
    "save_dir": "./checkpoints_optimized",
    "log_interval": 25,
    "eval_interval": 200
  },
  "data": {
    "data_dir": "./data",
    "max_length": 512,
    "stride": null,
    "num_workers": 8
  },
  "device": "cuda",
  "seed": 42
}