You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
in the webapp, when adding files, users might think that they are uploading data to the server, that it is not kept locally. labelling the button stating that would help ("connecting data" or "direct the model to your data"), with a tip/long description explaing it more ("Your data always stays on your computer. Only the AI model is downloaded and uploaded. Don't trust us? Try linking a large file and check your network usage with Task Manager on Windows or Activity Monitor on Mac"). better even would be to be explicit on the training process, such as having an animation or progress bar showing "downloading model from server", "training model in browser", "sharing model to server".
we can show the server node and maybe even where it's located for transparency (e.g. All of them would say "EPFML GitHub" at the moment), then the amount of users would be connected with a pipe to a cloud or something saying the amount of users instead of the textbox there is now, then there would be the user node which would show an animation of files going into it, staying there, an animation when the model is fetched from server to user node, animation for the model being trained and updated within the user node, then the upload of the model to server, all while the file icon is in the user node
The text was updated successfully, but these errors were encountered:
I very much agree with that, the current data form really looks like users are uploading their data. We could also add something like a lock icon and a mention of "Secure Form" (and make the form secure :) ) and something showing that their data has not been uploaded anywhere
Most won't be technical enough to check the code, but we could also hyperlink the parts on the repo that do the data linking when tfjs accesses data
Also maybe a link to a "How does it work" where we have a small explanation on federated learning and decentralized learning, or link to the information tab and have more explanation on the first information page
in the webapp, when adding files, users might think that they are uploading data to the server, that it is not kept locally. labelling the button stating that would help ("connecting data" or "direct the model to your data"), with a tip/long description explaing it more ("Your data always stays on your computer. Only the AI model is downloaded and uploaded. Don't trust us? Try linking a large file and check your network usage with Task Manager on Windows or Activity Monitor on Mac"). better even would be to be explicit on the training process, such as having an animation or progress bar showing "downloading model from server", "training model in browser", "sharing model to server".
animation idea from @EduardDurech:
The text was updated successfully, but these errors were encountered: