报错:param is too old, please regenerate

[lyp0413]

tensorflow2ncnn转出的模型，运行时遇到这个问题，希望有大神解答

[nihui]

generally, it is hard to convert tensorflow model to others. you'd better try caffe pytorch or mxnet ...
see #5

[suzaku]

I came across the same issue when loading files generated by caffe2ncnn.

The prototxt and caffemodel files were already upgraded using upgrade_net_proto_text and upgrade_net_proto_binary, but still ncnn complained that the param is too old.

[lyp0413]

@suzaku How did you load your model data to your code?

[suzaku]

@lyp0413

As documented in the wiki:

ncnn::Net net;
net.load_param("alexnet.param");
net.load_model("alexnet.bin");

With the parameters replaced with my own files.

[lyp0413]

@suzaku
check the first line in your .param.
Your model was not converted well,if it is not 7767517.