Larry Tsai

看到了。第二步的模型文件里面存了还有optimizer log之类的东西所以大了。https://github.com/nightsnack/YOLObile/blob/2494a48ff3591da3b400dcb3c531a77d22d510e4/train.py#L667 ，你改一下。这里只存model。部署手机需要compiler，compiler没有开源

256M有什么问题吗？这是存的.pt，pruning之后sparse的model里面0也是存下来的呀。在手机上运行的时候才会重新按照index-val的方式重新存储

Yes. Try it pelase.

Hi hhb1224, The first step includes 4 iterations of rho (in cfg/darknet_admm.yaml: --rou-num) each iteration uses a different rho value. Starting from 0.0001, the rou value increases by 10x each...

Sorry, I'm not familar with TX2. Perhaps you need a compiler. Please refer to the pinned issue. #7

是改了前几层的CSP connection的yolov4