{ "d_ff": 3072, "d_kv": 64, "d_model": 768, "dropout_rate": 0.1, "finetuning_task": null, "initializer_factor": 1.0, "is_decoder": false, "layer_norm_epsilon": 1e-06, "n_positions": 512, "num_heads": 12, "num_labels": 2, "num_layers": 12, "output_attentions": false, "output_hidden_states": false, "output_past": true, "pruned_heads": {}, "relative_attention_num_buckets": 32, "torchscript": false, "use_bfloat16": false, "vocab_size": 32128 }