{ "architectures": [ "NystromformerForMaskedLM" ], "attention_probs_dropout_prob": 0.1, "bos_token_id": 0, "conv_kernel_size": 65, "eos_token_id": 2, "hidden_act": "gelu_new", "hidden_dropout_prob": 0.1, "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "inv_coeff_init_option": false, "layer_norm_eps": 1e-05, "max_position_embeddings": 1024, "model_type": "nystromformer", "num_attention_heads": 12, "num_hidden_layers": 12, "num_landmarks": 64, "pad_token_id": 1, "segment_means_seq_len": 64, "torch_dtype": "float32", "transformers_version": "4.18.0.dev0", "type_vocab_size": 2, "vocab_size": 30000 }