🫘 Add set_seed() call in GRPO to ensure unique seed for each process

[qgallouedec]

Fixes the issue described in #2776 (comment)

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[mark-mcl]

this seems to also affect sampling, tried patching this and I get a different prompt on each device now

[kiddj]

thanks! does using a different seed for each process affect training?

[qgallouedec]

True @kiddj!! Fixed in 95fcfeb

[lewtun]

Nice fix, and could possibly benefit from a regression test to check completions are indeed unique across devices (ignore if you want to go fast though!)

[qgallouedec]

I've checked it locally, and yes it's now working as expected.
I'll see in a follow up PR if I can add a test for this