Closed lllkiller-jlll closed 3 years ago
Can you try following: vai_c_caffe --prototxt quantize/deploy.prototxt --caffemodel quantize/deploy.caffemodel --output_dir compile/ --net_name tmp --arch /opt/vitis_ai/compiler/arch/DPUCADX8G/ALVEO/arch.json --options "{'quant_cfgfile':'quantize/quantize_info.txt'}"
Are you using the latest version of run.py?
Hi sumitn.
I succeed convert custom model for DPUv1.
Thank you very much!
I'll close this issue.
Hi.
I'm trying to use own model on Alveo U250 accelerator card.
I saw (https://github.com/Xilinx/Vitis-AI/blob/master/alveo/examples/caffe/run.py) and I made deploy.prototxt, deploy.caffemodel, and quantize_info.txt successfully.
Finally, I input vai_c_caffe command but I faced the error message below:
`
Traceback (most recent call last): File "/opt/vitis_ai/conda/envs/vitis-ai-caffe/bin/vai_c_caffe", line 196, in
compiler = VAI_Caffe_Frontend(args)
File "/opt/vitis_ai/conda/envs/vitis-ai-caffe/bin/vai_c_caffe", line 93, in init
u = eval(args.options) if args.options else {}
File "", line 1, in
NameError: name 'quant_cfgfile' is not defined
`
My command was :
vai_c_caffe --prototxt quantize/deploy.prototxt --caffemodel quantize/deploy.caffemodel --output_dir compile/ --net_name tmp --arch /opt/vitis_ai/compiler/arch/dpuv1/ALVEO/ALVEO.json --options {"quant_cfgfile":"quantize/quantize_info.txt"}
What did I miss?
Thanks.