Dose your evaluation code support data parallelism?

[selmiss]

Dose your evaluation code support data parallelism? I couldn't find any module to mteb evaluation that use DP or DDP

[Muennighoff]

Yes it automatically uses DP via this line of code

gritlm/gritlm/gritlm.py

Line 75 in b89fdef

self.model = torch.nn.DataParallel(self.model)

[selmiss]

Thank you. And by the code we can only use DP or PP, can't mix using these two?

[Muennighoff]

Thank you. And by the code we can only use DP or PP, can't mix using these two?

Yeah only one or the other for mteb evaluation; I'm not sure if transformers supports mixing them