Spaces:
Running
on
CPU Upgrade
tons of models are stuck in the queue
One model decided to take an extended vacation inside the queue, and now 20+ of its friends are stuck behind it, honking their virtual horns. Looks like our queue needs a traffic cop...or maybe just a bigger eject button.
please reset the queue, we don't want these models stuck.
Feels like an Ethereum transaction.
+1
Most likely gpus are busy on different task, or they can't fix cuda problem they have with some models. And where get information wtf is going on is unknown.
Hi!
We are running on the spare cycle of the cluster, which means that if cool research experiments are running (like open r1, open deepresearch, smolLMs, etc), the leaderboard just waits for the more important xps to run. This is a normal behavior and documented
And it's unrelated to previous model evaluations, all evals are independant
All good! Thanks for the update!