{ "vocab_size": 128256, "tie_word_embeddings": false, "max_position_embeddings": 8192, "rope_theta": 500000.0, "rope_scaling": null, "hidden_size": 4096, "num_hidden_layers": 32, "rms_norm_eps": 1e-05, "num_attention_heads": 32, "num_key_value_heads": 8, "intermediate_size": 14336, "hidden_act": "silu", "initializer_range": 0.02, "model_type": "llama", "architectures": [ "LlamaForCausalLM" ] }