PP-OCRv5_server_det训练自己的模型，转换为 .nb 部署，运行时报错，提示需要很大很大的内存，请问怎么处理

[GaoMeng-2021]

以 PP-OCRv5_server_det 为基础，添加了自己的标记，进行训练。在PC上，可以正常识别图像。
使用指令转换为 .nb 模型，指令为：
python tools/export_model.py -c "./configs/det/PP-OCRv5/PP-OCRv5_server_det_new01.yml" ^
-o Global.export_with_pir=False Global.pretrained_model="./output/ch_PP-OCRv5_REC_new01/best_model/model.pdparams" ^
Global.save_inference_dir="./inference_model/det_v5_server_new01/best_model/"

paddle_lite_opt_zyg --model_file=./inference.pdmodel
--param_file=./inference.pdiparams
--optimize_out=optimized_model_det
--optimize_out_type=naive_buffer
--valid_targets=arm

转换为 optimized_model_det.nb 文件，放入板子中，启动识别操作，提示：
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1275 Setup] ARM multiprocessors name: HARDWARE : ROCKCHIP RK3568 EVB1 DDR4 V10 BOARD

[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1276 Setup] ARM multiprocessors number: 4
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1278 Setup] ARM multiprocessors ID: 0, max freq: 1992, min freq: 1992, cluster ID: 0, CPU ARCH: A55
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1278 Setup] ARM multiprocessors ID: 1, max freq: 1992, min freq: 1992, cluster ID: 0, CPU ARCH: A55
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1278 Setup] ARM multiprocessors ID: 2, max freq: 1992, min freq: 1992, cluster ID: 0, CPU ARCH: A55
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1278 Setup] ARM multiprocessors ID: 3, max freq: 1992, min freq: 1992, cluster ID: 0, CPU ARCH: A55
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1284 Setup] L1 DataCache size is:
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1286 Setup] 32 KB
[I 7/ 1 11:33: 2.886 ...addle-Lite-2.12/lite/core/device_info.cc:1286 Setup] 32 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1286 Setup] 32 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1286 Setup] 32 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1288 Setup] L2 Cache size is:
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1290 Setup] 512 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1290 Setup] 512 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1290 Setup] 512 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1290 Setup] 512 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1292 Setup] L3 Cache size is:
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1294 Setup] 0 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1294 Setup] 0 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1294 Setup] 0 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1294 Setup] 0 KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1296 Setup] Total memory: 1998392KB
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1297 Setup] SVE2 support: 0
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1298 Setup] SVE2 f32mm support: 0
[I 7/ 1 11:33: 2.887 ...addle-Lite-2.12/lite/core/device_info.cc:1299 Setup] SVE2 i8mm support: 0
[I 7/ 1 11:33: 3.325 ...e-Lite-2.12/lite/operators/squeeze_op.cc:124 AttachImpl] PaddleLiteV2.12 remove XShape OutputTensor for SqueezeOp.
[I 7/ 1 11:33: 3.534 ...e-Lite-2.12/lite/operators/squeeze_op.cc:124 AttachImpl] PaddleLiteV2.12 remove XShape OutputTensor for SqueezeOp.
[F 7/ 1 11:33:21.489 ....12/lite/backends/host/target_wrapper.cc:33 Malloc] Check failed: p: Error occurred in TargetWrapper::Malloc period: no enough for mallocing 18446744071803895808 bytes.
terminate called after throwing an instance of 'paddle::lite::PaddleLiteException'
what(): Paddle-Lite C++ Exception:
[F 7/ 1 11:33:21.489 ....12/lite/backends/host/target_wrapper.cc:33 Malloc] Check failed: p: Error occurred in TargetWrapper::Malloc period: no enough for mallocing 18446744071803895808 bytes.

按提示看，要求申请的内存为 18446744071803895808 bytes. 这个明显异常，请问要如何处理？

[huz1xuan]

你是通过什么转paddlelite的，用opt吗，paddleocr3.0后用json格式替代pdmodel格式后，是怎么正常使用opt转nb的呢