{ "d_model": 320, "in_model": 1280, "n_heads": 20, "intermediate_size": 960, "n_layers": 6, "norm_eps": 1e-05, "expr_bin": 51, "initializer_range": 0.02, "attention_out_bias": false, "norm_type": "layer_norm", "attention_dropout": 0.0, "mlp_dropout": 0.0, "mlp_in_proj": false, "traditional_mlp": false }