David Xu

This env got error "ERROR - mmdeploy.apis.pytorch2onnx.torch2onnx with Call id: 0 failed. exit." 2022-07-24 20:59:17,261 - mmdeploy - INFO - 2022-07-24 20:59:17,261 - mmdeploy - INFO - **********Environmental information********** 2022-07-24...

#780 It seems related to this. But this not work for me.

cuda 11.3, cudnn 8.2.0

(mmdeploy) D:\1-code\pytorch\mmdeploy>python tools/deploy.py configs\mmdet\detection\detection_tensorrt_dynamic-320x320-1344x1344.py D:\1-code\t_deploy-compare\configs\faster_rcnn\faster_rcnn_r50_fpn_1x_coc o.py D:\1-code\t_deploy-compare\models\faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth D:\1-code\t_deploy-compare\models\demo.jpg --work-dir D:\1-code\t_deploy-compare\models --device cuda:0 --dump-info D:\software\anaconda3\envs\mmdeploy\lib\site-packages\mmdet\datasets\utils.py:66: UserWarning: "ImageToTensor" pipeline is replaced by "DefaultFormatBundle" for batch inference. It is recommended to manuall y...

when debug the code, the following code run with error. ``` torch.onnx.export( patched_model, args, output_path, export_params=True, input_names=input_names, output_names=output_names, opset_version=opset_version, dynamic_axes=dynamic_axes, keep_initializers_as_inputs=keep_initializers_as_inputs, verbose=verbose) ```

> Netron 我试了，但是对于检测任务都只会出现下面独立模块，不知道你那边存储模型的时候是否做了特殊处理呢？ 

我的模型是用https://github.com/jwyang/faster-rcnn.pytorch 代码训练出来的，自己改了些东西，用Netron打开就是上面那样，我看了下Netron的源码，可惜electron不熟。。 你能帮忙提点意见嘛

看不到箭头，不知道数据流流向以及各个地方的输出维度。。

关键我用的是腾讯云的云主机。。会有影响嘛

这样啊，我的内存56G，估计硬盘不够，硬盘只有50G，我换个大点的硬盘试试，谢啦