{ "d_model": 1024, "d_ff": 4096, "d_kv": 64, "num_heads": 16, "num_layers": 24, "num_decoder_layers": 24, "dropout_rate": 0.0, "init_method_std": 0.001, "initializer_factor": 1.0, "layer_norm_epsilon": 1e-06, "max_position_embeddings": 512, "use_cache": false, "attn_scale": true }