continuous batching wins and losses

[tensorkite]

What worked for me was pretty simple. On GB200, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[deltapoint17]

I'd start with the simpler stack.

[routearch]

Appreciate the detail. I don't think your conclusion is far off. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.