human evals on small teams for Phi

[southpatch]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Phi and compare from there. Not glamorous advice, but repeatable baselines save a lot of time. At this point the real bottleneck is my patience.

[cudapeak]

Can you fill in a bit more context? Are you testing this on 7900 XTX or something else?