模型通过剪枝后与原yolov5s.pt的模型里的参数名称不一致吗

ZJU-lishuang / yolov5_prune

yolov5 prune，Support V2, V3, V4 and V6 versions of yolov5

Apache License 2.0

556 stars 136 forks source link

Open bingo789 opened 1 year ago

bingo789 commented 1 year ago

通过剪枝后再微调得出的模型，放到yolov5里导出onnx，结果报错 PyTorch: starting from best.pt with output shape (1, 25200, 85) (5.9 MB) 段错误 (核心已转储)

请问可以转成跟yolov5s.pt参数名一样吗？这样方便在原有的tensorrt加速模型上修改，不用重新写一套