How can I store files and model weights from Google Colab to my Google Drive?

Other Questions Google drive
1

I’m working with Google Colab and need to save my generated images and trained model weights to Google Drive instead of the default temporary storage. Right now when I check my current working directory, it shows /content which is just temporary storage.

import os
current_path = os.getcwd()
print(current_path)
# Output: /content

Is there a way to mount Google Drive or change the save location so my files don’t get lost when the runtime disconnects? I want to make sure my work is permanently saved to external storage like Google Drive or access local drives if possible.

2

here’s another trick - use symbolic links after mounting your drive. once you run drive.mount('/content/drive'), create a symlink with !ln -s /content/drive/MyDrive/colab_files /content/files. now you don’t have to remember that long path every time. just save to /content/files/ and it goes straight to your drive automatically.

3

You can save files and model weights to Google Drive from Colab using the drive module. Import it with from google.colab import drive, then mount using drive.mount('/content/drive'). Be sure to authenticate your Google account during this process. Once mounted, your Drive files will be accessible under /content/drive/MyDrive/. It’s wise to create a dedicated folder in your Drive for your Colab projects. When saving your model weights, use the complete path like /content/drive/MyDrive/my_project/model.pth to ensure that you don’t lose any files when the session disconnects. Keep in mind that saving to Drive might take more time compared to local storage.

4

I’d set up automatic syncing to prevent data loss. Mount your Drive normally, then create a backup function that saves your work periodically. You can also update your existing save commands - just change the file paths to point to your mounted Drive location. For example, with torch.save() or model.save(), change the path parameter to /content/drive/MyDrive/your_folder/. Watch out for upload speed though - larger model files can take several minutes to sync. I save checkpoints at regular intervals during training instead of waiting until the end, especially for longer sessions where connection drops are more likely.