{ "architectures": [ "MurzikForCausalLM" ], "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_murzik.MurzikConfig", "AutoModel": "modeling_murzik.MurzikForCausalLM", "AutoModelForCausalLM": "modeling_murzik.MurzikForCausalLM", "AutoTokenizer": "tokenization_murzik.MurzikTokenizer" }, "bos_token_id": 7, "dtype": "bfloat16", "eos_token_id": 8, "head_dim": 128, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 20480, "max_position_embeddings": 8192, "model_type": "murzik", "num_attention_heads": 40, "num_hidden_layers": 32, "num_key_value_heads": 8, "pad_token_id": 6, "rms_norm_eps": 1e-06, "rope_theta": 1000000.0, "tie_word_embeddings": true, "transformers_version": "5.6.0", "use_cache": false, "use_qk_norm": true, "vocab_size": 128256 }