Testing a fix for the dependabot auto_updates

research/gemini/README.md

@@ -8,7 +8,7 @@ As the training data associated with the GEMINI experiments is private and sensi
 
 **Note**: These scripts are not runnable outside of the GEMINI environment. They are for exposition purposes and to aid reproducibility for those with access to the GEMINI HPC system and data.
 
-You can find the scripts and commands to run the experiments in each algorithm folder. This codebase is built using two different versions of the FL4Health library, therefore, there is a `requirements1.txt` that should be activated for all the methods except `FedPer`, `PerFCL`, `Ditto`, and `MOON`. The modules specified in `requirements2.txt` should be activated to run `FedPer`, `PerFCL`, `Ditto`, and `MOON`.
+You can find the scripts and commands to run the experiments in each algorithm folder. This codebase is built using two different versions of the FL4Health library, therefore, there is a `first_requirements.txt` that should be activated for all the methods except `FedPer`, `PerFCL`, `Ditto`, and `MOON`. The modules specified in `second_requirements.txt` should be activated to run `FedPer`, `PerFCL`, `Ditto`, and `MOON`.
 
 ## Delirium extreme heterogeneity experiments
 The experiments, by default, use naturally heterogeneous data splits. To perform extreme heterogeneity experiments, use `300` as the model's first layer size instead of `8093` when defining the model. This change can be easily applied by setting this parameter in model constructor. Don't forget to apply the change in both `client.py` and `server.py`. Also, the path to the data source should be adjusted to `data_path = Path("heterogeneous_data")`