Li Lele
Li Lele
> 设置成你想放置的文件夹即可，format： experimentWorkingDirectory: "xx" xx最好是绝对路径 这个方法我尝试过，没有解决问题，在终端执行py文件，可以看到nni返回了中间结果，但就是生成不了trial.log文件，更换nni版本也不行，以前我使用nni==2.5.0是可以的，现在不行了
> > > > > > 设置成你想放置的文件夹即可，format： experimentWorkingDirectory: "xx" xx最好是绝对路径 > > > > > > 这个方法我尝试过，没有解决问题，在终端执行py文件，可以看到nni返回了中间结果，但就是生成不了trial.log文件，更换nni版本也不行，以前我使用nni==2.5.0是可以的，现在不行了 > > 可能还是路径问题吧，默认情况下，nni-experiments这个文件夹不在content下，我这个问题可能也是误打误撞解决了，抱歉啊家人！这个是我的config.yml，我每次都是路径切换到train_reg.py所在文件夹，然后在ipynb文件中运行： searchSpaceFile: space.json experimentName: pyg experimentWorkingDirectory: "/content/drive/MyDrive/exp" trialCommand: python train_reg.py trialGpuNumber:...
I also have this error ```puthon import sys print(sys.version,sys.platform) 3.9.7 (default, Sep 16 2021, 16:59:28) [MSC v.1916 64 bit (AMD64)] win32 ``` ERROR: Could not find a version that satisfies...
同问丫
me too! I adjust 3h, it still don't show! only this way!
> Hi, the reason are is that the project is working with multiple different image sizes and sequence lengths. > > It seems like the one of the largest combinations...
You have the same problem as I do, and my suggested solution is to reduce the grid and increase the number of layers. You can try it.
When there are more grids, it does make the network more expressive (more diverse activation functions), but sometimes depth is necessary (for complex problems). I think that the trainable activation...