{
    "architectures": [
        "LlamaForCausalLM"
    ],
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": 128000,
    "eos_token_id": [
        128001,
        128008,
        128009
    ],
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 8192,
    "initializer_range": 0.02,
    "intermediate_size": 28672,
    "max_position_embeddings": 131072,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 64,
    "num_hidden_layers": 80,
    "num_key_value_heads": 8,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-05,
    "rope_scaling": {
        "factor": 8.0,
        "high_freq_factor": 4.0,
        "low_freq_factor": 1.0,
        "original_max_position_embeddings": 8192,
        "rope_type": "llama3"
    },
    "rope_theta": 500000.0,
    "tie_word_embeddings": false,
    "torch_dtype": "float16",
    "transformers_version": "4.47.0.dev0",
    "use_cache": true,
    "vocab_size": 128256,
    "quantize": "w8a8",
    "quantization_config": {
        "group_size": 0,
        "kv_quant_type": null,
        "fa_quant_type": null,
        "w_bit": 8,
        "a_bit": 8,
        "dev_type": "npu",
        "fraction": 0.01,
        "act_method": 3,
        "co_sparse": false,
        "anti_method": "m3",
        "disable_level": "L5",
        "do_smooth": false,
        "use_sigma": false,
        "sigma_factor": 3.0,
        "is_lowbit": false,
        "mm_tensor": false,
        "w_sym": true,
        "open_outlier": true,
        "is_dynamic": false
    }
}