{
    "error_type": "memory_error",
    "description": "内存不足导致推理失败",
    "error_patterns": [
        "OOM: out of memory during tensor allocation",
        "allocate failed: insufficient memory for intermediate tensors",
        "memory allocation error: cannot allocate 256MB buffer",
        "memory error: GPU memory exhausted",
        "cannot allocate memory for model weights"
    ],
    "diagnosis": {
        "cause": "模型过大或输入数据batch size过大，超出可用内存限制",
        "check_points": [
            "模型文件大小是否超过内存容量",
            "输入数据的batch size大小",
            "是否使用了量化或FP16压缩",
            "运行设备(GPU/NPU)的内存/显存大小"
        ]
    },
    "suggestions": [
        "启用INT8量化重新转换模型以减小模型大小",
        "减小输入数据的batch size",
        "使用converter_lite的--fp16=on减小模型体积",
        "切换到CPU运行避免GPU显存限制"
    ],
    "example_log": "[ERROR] OOM: out of memory during inference. Failed to allocate tensor buffer of size 256MB. Available memory: 128MB. Memory exhausted during layer 'conv_block_5'."
}