{ "vocab_size": 45093, "d_model": 1280, "n_heads": 20, "n_layers": 24, "d_ff": 5120, "max_seq_len": 300, "dropout": 0.1, "learning_rate": 0.0002, "weight_decay": 0.01, "warmup_steps": 1000, "gradient_clip_val": 1.0, "bos_token": 0, "eos_token": 1, "pad_token": 2, "val_loss": 2.6841174858372385 }