Open Sybil-Huang opened 3 years ago
I concatenated the dataset and put all data points into the matrix. Here are the examples: [datasets]https://drive.google.com/drive/folders/13L3Wx9dmY_WUEcDmR_SQdwISXrNdsS5_?usp=sharing)
Thanks for your answering and it really helps. I'm sorry to disturb you again for another problem under the non-iid condition. When I ran the code and set the 'iid: false' in file 'config.yaml', it reported an error https://imgtu.com/i/6EsQ3j. Is it related to the paths of '_user_withdata' and '_full_weightdir' in file 'config.yaml' since I didn't have file '_user_dataidx_map0.10.dat' and '_fullinit.pth' ? How can I solve this problem? I'm looking forward to your reply.
For the "pth" file, it is the pytorch model state_dict. You can either use some pre-trained model or a carefully-initialized model. For "user_with_data", it is a map between user_id and training data point id. You can easily build one based on your own setting.
It is really nice of you to share such a helpful and clear repo. I tried to run this code but it reported an error that files "train.dat" and "test.dat" couldn't be found. Could you please share your data files if possible?