Traditionally, downloading large files onto a virtual server (like RunPod) can be a cumbersome process using command line tools like Runpodctl or Croc. To circumvent this difficulty and save valuable time, a user interface (UI) based solution that eliminates the need for coding has been developed.
The created tool is a Jupyter Notebook that requires the following two components:
1. Google Drive: This is the platform where model files or any other large files need to be stationed.
2. Jupyter Notebook: This downloadable notebook (available on the website) aids in transferring files from Google Drive to RunPod.
Procedures on Google Drive:
Firstly, users must log in to Google Drive and upload the model file or any sizeable data files onto a chosen folder. Then, the file needs to be selected and shared via a link. It’s important to ensure that permissions are set to ‘anyone with the link can view.’ The link should then be copied for future use.
The same set of steps needs to be repeated for each file that is desired to be downloaded and utilized on RunPod. Anonymity of users is given utmost importance during this process as the links are exclusively accessible by the user.
After usage, it is advised to un-share the files for security purposes.
Steps for Jupyter Notebook:
The process begins by uploading the Jupyter Notebook file (.ipynb) to a chosen folder. Users then need to open and run the notebook.
The tool, GDown, should be installed. Next, a unique URL (a shared link from Google Drive) and output for each file need to be updated in their respective cells. The output is the complete file path where the file will be saved, for example, /workspace/albedobaseXL_v20.safetensors.
The notebook provides two cells as default, but they can be duplicated further if numerous files need to be downloaded. The downloaded file can be found in the designated file path. The entire procedure accomplishes downloading in mere minutes which is a significant enhancement compared to traditional methods.
It is recommended to download and locally save an updated copy of the notebook to eliminate the repetitive URLs update for each usage. This modified file can then be uploaded to RunPod and utilized for swift downloads.
This tool supports all Jupyter lap installs making it universally compatible, hence it can operate on other service providers as well.
To support further site developments and content creation, users are encouraged to buy a product, give a Ko-fi, or subscribe. Affiliate or referral links on the website also contribute to site support without any additional cost to users.
