what are you all running this week on GB200

[ggufwire22]

A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at NVFP4. Not glamorous advice, but repeatable baselines save a lot of time.

[thetaember38]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Kimi and compare from there.

[tensorkite]

Can you fill in a bit more context? What model, what runtime, and roughly what context size?