When will higher allow the use of DDP (distributed data parallel)?

[brando90]

I remember from this post #99 that DDP seemed to be a feature that was in process of dev for higher (I remember @albanD was tagged in that post).

I was curious on the progress for that?

Thanks!

related: #98

[kamalojasv181]

Any updates when this will happen?

[kamalojasv181]

@brando90 Is it possible for me to just use your stateless model and not your optimizers and fast net API, and do distributed training? Like I calculate the parameters of the fnet manually by subtracting the parameters by the gradient, and using my updated params to meta learn? Will this work for now?

[brando90]

I don't know. But if it works let me know! Feel free to use my code! Should be available online I hope.

…

On Feb 1, 2022, at 12:32 AM, OJASV Kamal ***@***.***> wrote: @brando90 <https://github.com/brando90> Is it possible for me to just use your stateless model and not your optimizers and fast net API, and do distributed training? Like I calculate the parameters of the fnet manually by subtracting the parameters by the gradient, and using my updated params to meta learn? Will this work for now? — Reply to this email directly, view it on GitHub <#116 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAOE6LV6XKZJR4SRC4S7I3TUY55AFANCNFSM5E5YVPOA>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.

[kamalojasv181]

@brando90 Im sorry, which code? I used the pronoun "your" to refer to the higher code.