automl
Document better how to use offline
Just some suggestion:-
Since TabPFN uses data large pre-trained weights and possibly sample datasets that are usually downloaded from remote servers, we can download the files ( dependencies ,model weights, and datasets locally ) so they can be accessed without the internet.
Download the pre-requisites using
pip download -r requirements.txt -d offline_packages/
and when offline install them locally using
pip install --no-index --find-links=offline_packages/ -r requirements.txt
The pre-trained model is likely stored in an online repository and if not cached it will be downloaded from internet when TabPFN will be in use. Moreover if a package isn’t installed, pip may try to fetch it from PyPI. Pre-downloading packages avoids this.
There arn't alot of problems of using it offline, just some minor fixed have to be done.
I don't see requirements.txt on the repo
Trying to find a way to work with this model offline
