Closed GiuseppeSerra93 closed 7 months ago
Also, I am confused about the server_ss
parameter, could you please clarify what it does and how to use it? Many thanks!
Hi Guiseppe,
You are correct about the syn_size, and its values should be 32. I think 128 is a problem with the code.
Also, server_ss is the server sampling size and is responsible for the batch size in training the generative model itself.
Best, Sara
Thanks for the clarification! So, for server_ss
, 128 is correct?
You are welcome and yes.
On Mon, Apr 29, 2024 at 12:05 AM Giuseppe Serra @.***> wrote:
Thanks for the clarification! So, for server_ss, 128 is correct?
— Reply to this email directly, view it on GitHub https://github.com/SaraBabakN/MFCL-NeurIPS23/issues/4#issuecomment-2082020286, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANA6MGZR3VDRMXL5ZN5XVWLY7XWMRAVCNFSM6AAAAABGNQODI2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBSGAZDAMRYGY . You are receiving this because you commented.Message ID: @.***>
Hello,
I was checking something when I realised the
syn_size
parameter is set to 128. From my understanding, this hyperparameter controls the size of the synthetic mini-batch. However – according to Section F of the appendix – this value is set to 32 for all the experiments. Am I missing something? Many thanks for any clarification!Best, Giuseppe