v4tiny_pt_quant

Quantization for yolo with xilinx/vitis-ai-pytorch

Tool

• Quantization tool: xilinx/vitis-ai:latest, vitis-ai-pytorch
• Environment setup: Vitis-AI 1.3

Model

• Model: yolov4-tiny pruned with this project or other yolo models
• Dataset: coco2017, with darknet format
• Input format: .cfg + .pt/.weights

Errors

• Error unfixed: aten_op 'view' parse failed(the layout of activation of reshape is ambiguous) when run with python quant.py --quant_mode calib