onnx-tf convert onnx to pb format successfully, but no output layer

[noending]

when i use "onnx-tf convert -t tf -i squeeze1-11.onnx -o squeeze1-11.pb" to convert squeezent from onnx to pd format, i got pb file without any errors,

but when use netron to see the onnx and pb file, pb without output

can anyone tell me how to solve this issue, many thx. By the way squeezenet was trained by pytorch1.0, I convert it from pytorch pth to onnx and then to pb, trying to deploy on the mobile device.

[ftaralle]

Hi,
I got the same "issue" converting pytorch -> onnx -> tensorflow -> tensorflow-lite
In fact, in netron, you can click on the last "Reshape" block, to show the information pannel.
it seems we can use the name provided in NODE PROPERTIES
to be noted that I havent tried the resulting tensorflow model yet.

[fumihwh]

@noending Is Reshape output's name output1 in pb?

[noending]

Hi,
I got the same "issue" converting pytorch -> onnx -> tensorflow -> tensorflow-lite
In fact, in netron, you can click on the last "Reshape" block, to show the information pannel.
it seems we can use the name provided in NODE PROPERTIES
to be noted that I havent tried the resulting tensorflow model yet.
@ftaralle many thx, i've retrained nn in tf and coverted it to tflite.

[noending]

@noending Is Reshape output's name output1 in pb?

@fumihwh 当然不是啊， 已经弃坑， 用tf重新训练了