Tired of trying to use Runpodctl or Croc or Wget to download large checkpoint files or worse uploading via the Jupyter file browser which will take lot of time to upload. When you are running and using Cloud GPU like RunPod, time is literally money and last thing you want to do is wait for uploads and downloads to finish on your virtual server.

Many RunPod template refer you to command line tools which you need to use to download from remotely hosted files. However my solution is UI based and requires no coding to be done.

I have build a Jupyter Notebook which you can download large files onto your RunPod quickly using two things:

  • Google Drive – this is where you need to host your model/large files
  • Jupyter Notebook – my notebook which you can download below
Runpod Jupyter Notebook (98 downloads )

Google Drive Setup

Login to your Google drive and upload the model/large files onto your desired folder.

  • Select the file and click on Share
    • Make sure you give “Anyone with the link” access permissions as Viewer.
    • Copy the link generated.

Repeat the above steps once per file that you want to download and use on RunPod. Of course you alone have these links so nobody else would be able to figure out the link on their own. However I suggest that when you don’t need you still un-share these files.

The Link generated would look something like this: https://drive.google.com/file/d/1rPu7sdiukefweoiullkjsdfLDaqX/view?usp=sharing

Jupyter Notebook

Upload the Jupyter Notebook file (.ipynb) to your desired folder, in my case this is /workspace/. Next double click on the Notebook file to Launch it.

Run the Install the GDown cell. You will see similar results below showing the installation is complete and requirements are met or installed.

Next update the URL and output in each of the cell as needed. The URL is the shared link from Google Drive that you should copy and paste. Output is the full file path where you will saving the file eg. /workspace/albedobaseXL_v20.safetensors

The notebook provides two cells by default but you can duplicate them further depending upon how many files you need to download. I generally use 1 cell per file so I can selectively run and download each file.

You can see the AlbedobaseXL_v20.safetensors was downloaded from Google Drive is under 2 mins which is amazing and the small LoRA file add_details.safetensors in under 2 seconds.

Once you have updated the notebook I suggest that you download this updated copy and keep it on your computer so you don’t have update and add the URLs each time. You can simply upload the Notebook file to you RunPod and run it to download all the models.

Conclusion

Hope you found this post & my notebook useful in your worklfow and will enjoy using it on your RunPod instance. The notebook is compatible with any JUPYTER lap install so you can run it on other service providers like.

If you’d like to support our site please consider buying us a Ko-fi, grab a product or subscribe.

Some of the links on our site are affiliate or referral links. These help support our site, if you use these links and make a purchase it doesn’t cost you more just helps us keep creating more content.

If you'd like to support our site please consider buying us a Ko-fi, grab a product or subscribe.