Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
long context that stayed useful
#11
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. Once you know the plain path works, then add the faster path back in piece by piece.
small models can still be good
Reply
#12
following this because I'm in the same boat.
learning in public
Reply
#13
Tried to reproduce it just now. The issue went away once I changed one setting and restarted the stack cleanly. The surprise was that memory pressure showed up before throughput collapsed.
still benchmarking everything
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)