small win thread

[cacheplane98]

Spent some time on this today and wanted to compare notes.

I'm comparing two approaches here: keep it simple on GB200, or squeeze more out of the stack and deal with the complexity.

Interested in tradeoffs more than fan loyalty.

[thetagrid15]

I'd approach it in a boring order. If you're on PyTorch, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

[deltastation]

Good point. The pattern around embeddings has been pretty similar for me.

[marbletrack]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Gemma and compare from there. Once you know the plain path works, then add the faster path back in piece by piece.

[ggufwire22]

Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly. I expected the model choice to matter more than it did.

[rocmplane]

Appreciate the detail. That matches what I saw on 7900 XTX. The biggest change for me was simplifying the stack instead of adding one more optimization layer.