pytorch-caffe2-aws-lambda
pytorch-caffe2-aws-lambda copied to clipboard
Error when using the precompiled 'deps.zip' from this repo
Hello and thanks for this amazing piece of work,
When trying to use your deps.zip in an AWS lambda deployment (with the same model you suggest in your tutorial in an S3 bucket), Lambda execution fails with the following error:
{
WARNING: "This caffe2 python run does not have GPU support. Will run in CPU only mode."
WARNING: "Debug message: No module named caffe2_pybind11_state_gpu"
ERROR: "The graph does not have an ir_version set properly"
}
I assume this may be an issue with the model? Any idea?
Thanks!
I've managed to pass this by essentially using v1.4 (latest, instead of 0.1 as in your write-up) of ONNX, but in this case the results of the pretrained network are empty:
Outputs()
when it should be:
Outputs(x,y,z)
Once again, any help would be greatly appreciated :)
Thanks!