DavdGao
**Describe the bug** As the title says, the current number of batches and epochs are calculated for each split as follows: ``` ... # Process training data if self.train_data is...
As the title says, since the dataset is loaded before both `fedrunner.run()` and the setup of clients, the `config_per_client.yaml` doesn't work for the dataset, such as different batch size.
Maybe FS should allow to stop the training automatically if the `test_loss` is `NAN`, which is common during the hyper-parameter searching.
As the title says, we should solve the conflicts within the configs among different algorithms before calling `client.run()` or `server.run()`.
## Target - We are planning to add a consultation stage before training for differential privacy and other algorithms. - We decompose it into several PRs to avoid large modification...
## Our Target - We are planning to add a consultation stage before training for differential privacy. - We decompose it into several PRs to avoid large modification at one...
Main modifications: - Add functional mechanism for differential privacy, including: - a trainer named `fmtrainer` - two models for linear regression and logistic regression in funcational mechanism - Support gaussian...
- I'm trying to use grpc in my android project. - I have generated my java class by protoc (version: libprotoc 3.21.7). - Here is my proto file: ``` syntax...
## Target I want to build CLBlast for android app with JNI. ## Question I am cross-compiling clblast for android with the following scripts. - **Q1.** CLBlast requires OpenCL library,...
## Description Add a prompt tuning module, which provides functions: - System prompt generation (with few-shot examples) - System prompt tuning ## Checklist Please check the following items before code...