{ "activation_function": "gelu_new", "architectures": [ "SubSegGPTForCausalLM" ], "attention_dropout": 0.1, "bos_token_id": 1, "char_decoder_hidden_size": 256, "char_decoder_num_layers": 1, "char_embed_dim": 128, "dropout": 0.1, "dtype": "float32", "eos_token_id": 2, "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "layer_norm_eps": 1e-05, "lex_vocab_size": 20001, "max_position_embeddings": 1024, "max_seg_len": 5, "model_type": "subseggpt", "n_alpha": 263, "num_attention_heads": 12, "num_hidden_layers": 12, "pad_token_id": 0, "seg_end_token_id": 3, "transformers_version": "4.57.1", "vocab_size": 416 }