search

Xilinx / Vitis-AI

Vitis AI is Xilinx’s development stack for AI inference on Xilinx hardware platforms, including both edge devices and Alveo cards.
https://www.xilinx.com/ai
Apache License 2.0
1.47k stars 630 forks source link

export xmodel failed when use vai_q_pytorch #884

Closed JackzhangLOVEcode closed 2 years ago

JackzhangLOVEcode commented 2 years ago

hey, guys, i have completed calib mode, but when run test mode, TypeError come up, here are information: [VAIQ_NOTE]: =>Converting to xmodel ... Traceback (most recent call last): File "quantize_zc.py", line 147, in run_main() File "quantize_zc.py", line 140, in run_main quantize(args.build_dir,args.quant_mode,args.batchsize) File "quantize_zc.py", line 95, in quantize quantizer_en.export_xmodel(deploy_check=False, output_dir=quant_model_en) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/apis.py", line 127, in export_xmodel self.processor.export_xmodel(output_dir, deploy_check) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/qproc/base.py", line 248, in export_xmodel dump_xmodel(output_dir, deploy_check, self._lstm_app) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/qproc/base.py", line 321, in dump_xmodel deploy_graphs = get_deploy_graph_list(quantizer.quant_model, quantizer.Nndctgraph) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/qproc/utils.py", line 276, in get_deploy_graph_list g_optmizer = DevGraphOptimizer(nndct_graph) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/compile/deploy_optimizer.py", line 40, in init self._dev_graph.clone_from(nndct_graph) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/nndct_graph/base_graph.py", line 105, in clone_from self._top_block.clone_from(src_graph.block, local_map, converted_nodes) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/nndct_graph/base_block.py", line 47, in clone_from self.append_node(self.owning_graph.create_node_from(node, local_map, converted_nodes)) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/nndct_graph/base_graph.py", line 132, in create_node_from node.clone_from(src_node, local_map) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/nndct_graph/base_node.py", line 119, in clone_from self.op.clone_from(src_node.op, local_map) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/nndct_shared/nndct_graph/base_operator.py", line 221, in clone_from setattr(self, config, new_value) File "/opt/vitis_ai/conda/envs/vitis-ai-pytorch/lib/python3.7/site-packages/pytorch_nndct/parse/torch_op_def.py", line 203, in padding self.set_attr(self.AttrName.PAD, [value[0], value[0]]) TypeError: 'int' object is not subscriptable

i have no idea about the error, anyone can help? thank you for any advise

wangxd-xlnx commented 2 years ago

Hi @JackzhangLOVEcode

Could you provide some more details such as which model did you use and if you use Vitis-AI docker or your environment info?

JackzhangLOVEcode commented 2 years ago

Hi @JackzhangLOVEcode

Could you provide some more details such as which model did you use and if you use Vitis-AI docker or your environment info?

thanks for reply, I have solved this problem, the error is caused by the one-dimensional convolution function, I have changed it to two-dimensional convolution function.

zyt1024 commented 1 year ago

Hi @JackzhangLOVEcode Could you provide some more details such as which model did you use and if you use Vitis-AI docker or your environment info?

thanks for reply, I have solved this problem, the error is caused by the one-dimensional convolution function, I have changed it to two-dimensional convolution function.

Hello, I also encountered this problem in my network where I used Conv1D, but I'm using Vitis-AI 2.5. I would like to know if Conv1D is not supported in Vitis-AI ？

zyt1024 commented 9 months ago

Hi @JackzhangLOVEcode Could you provide some more details such as which model did you use and if you use Vitis-AI docker or your environment info?

thanks for reply, I have solved this problem, the error is caused by the one-dimensional convolution function, I have changed it to two-dimensional convolution function.

Hello, I also encountered this problem in my network where I used Conv1D, but I'm using Vitis-AI 2.5. I would like to know if Conv1D is not supported in Vitis-AI ？

@JackzhangLOVEcode Hello, I also encountered this problem in my network where I used Conv1D, but I'm using Vitis-AI 2.5. I would like to know if Conv1D is not supported in Vitis-AI ？