{
    "accum_grad": 1,
    "att_dropout_rate": 0.0,
    "att_unit": 512,
    "backend": "pytorch",
    "beta1": 0.9,
    "beta2": 0.999,
    "config2": null,
    "config3": null,
    "debugmode": 1,
    "dropout_rate": 0.0,
    "dump_hdf5_path": null,
    "early_stop_criterion": "validation/main/loss",
    "emb_dropout_rate": 0.0,
    "embed_unit": 128,
    "epoch": 50,
    "gradclip": 1.0,
    "head": 8,
    "layer": 16,
    "lr": 0.0008,
    "lr_cosine_total": 100000,
    "lr_cosine_warmup": 1000,
    "maxlen": 60,
    "model_module": "transformer",
    "n_vocab": 5049,
    "ngpu": 8,
    "opt": "adam",
    "patience": 0,
    "pos_enc": "none",
    "report_interval_iters": 100,
    "schedulers": [
        [
            "lr",
            "cosine"
        ]
    ],
    "seed": 1,
    "sortagrad": 0,
    "test_label": null,
    "tie_weights": false,
    "train_dtype": "float32",
    "unit": 2048,
    "verbose": 1,
    "weight_decay": 0.0
}