{ "architectures": [ "UTUModel" ], "attention_bias": false, "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_utu.UTUConfig", "AutoModel": "modeling_utu.UTUModel", "AutoModelForCausalLM": "modeling_utu.UTUForCausalLM" }, "bos_token_id": 128000, "embedding_initializer_range": 0.02795084971874737, "eos_token_id": 128001, "head_dim": 64, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.013975424859373685, "intermediate_size": 8192, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "utu", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 32, "pretraining_tp": 1, "rms_norm_eps": 1e-08, "rope_scaling": null, "rope_theta": 1600000.0, "tie_word_embeddings": true, "tokenizer_class": "PreTrainedTokenizerFast", "torch_dtype": "float32", "transformers_version": "4.51.3", "use_cache": false, "vocab_size": 128256 }