I succesfully create a slim_model with the vitis-ai-optimizer_pytorch. However, if i try to to the quantization with this conda env, i get the folowing error :
File "/opt/vitis_ai/conda/envs/vitis-ai-optimizer_pytorch/lib/python3.7/site-packages/nndct_shared/compile/xgraph.py", line 24, in <module>
raise Exception('please install xir package')
In the reverse, i can't load the slim_model in the vitis-ai-pytorch as describe in the ug1333 because of the folowing error:
ImportError: cannot import name 'get_pruning_runner' from 'pytorch_nndct' (/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/__init__.py)
How can i load the pruned model for the quantization step? Is there any exemple?
I am using the "With Pruning API" for loading the slim_model.
Hello Xilinx,
Thank you for alowing me to play with your tool,
I succesfully create a slim_model with the vitis-ai-optimizer_pytorch. However, if i try to to the quantization with this conda env, i get the folowing error :
In the reverse, i can't load the slim_model in the vitis-ai-pytorch as describe in the ug1333 because of the folowing error:
ImportError: cannot import name 'get_pruning_runner' from 'pytorch_nndct' (/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/__init__.py)
How can i load the pruned model for the quantization step? Is there any exemple?
I am using the "With Pruning API" for loading the slim_model.
The Without Pruning API doesent work for me. The model.json file does not fit the slim_model_from_state_dict_and_pruning_info function you provided.
Thanks