When I run 'python ./tools/train.py', I encounter the following problems. How should I solve it

[jiangliangliang2020]

[magnetmoment]

Python 3.6 is not compatible. You should try python 3.7. https://blog.csdn.net/weixin_44789039/article/details/120754136

[jiangliangliang2020]

Thanks

[Vegeta2020]

@jiangliangliang2020 It's caused by the copy.deepcopy() funtion, you may try to change the python version as suggested by @magnetmoment, or you may try to change this line to

model_ema = build_detector(cfg.model, train_cfg=cfg.train_cfg, test_cfg=cfg.test_cfg)
for param in model_ema.parameters():
    param.detach_()

as stated in this issue

[sixcomicpen]

hello！

[sixcomicpen]

Python 3.6 不兼容。你应该试试 python 3.7。 https://blog.csdn.net/weixin_44789039/article/details/120754136

I use version 3.7 of Python to display the following

[Vegeta2020]

@sixcomicpen How about installing spconv 1.0 or 1.1?

[sixcomicpen]

@sixcomicpen How about installing spconv 1.0 or 1.1? I got more errors with SPCONv 1.0,like this

[jiangliangliang2020]

Python 3.6 不兼容。你应该试试 python 3.7。 https://blog.csdn.net/weixin_44789039/article/details/120754136

I use version 3.7 of Python to display the following

spconv的版本跟python的版本要相匹配，你之前是在python3.6下编译的spconv，不能在python3.7下直接使用，应该重新编译

[sixcomicpen]

Python 3.6 不兼容。你应该试试 python 3.7。 https://blog.csdn.net/weixin_44789039/article/details/120754136

I use version 3.7 of Python to display the following

spconv的版本跟python的版本要相匹配，你之前是在python3.6下编译的spconv，不能在python3.7下直接使用，应该重新编译

这是我重新创建一个新的环境后，安装spconv1.0弹出来的错误