Description
For now, in utils.py the transforms are applied the same way as for the MNIST dataset, i.e. from a central source the data is loaded here ( this part can be left away in new approach), then split, and then transforms are applied here. Normally /data/server_simulation/distribute_data.py should be used to first create different folders (one for each worker) and then load the data from there. Once the distribution is implemented the code for loading and the transforms is already here. (slight changes might be necessary)
Description
For now, in utils.py the transforms are applied the same way as for the MNIST dataset, i.e. from a central source the data is loaded here ( this part can be left away in new approach), then split, and then transforms are applied here. Normally /data/server_simulation/distribute_data.py should be used to first create different folders (one for each worker) and then load the data from there. Once the distribution is implemented the code for loading and the transforms is already here. (slight changes might be necessary)