✨What's New - July Updates
Features
Users can now generate wireguard configuration from credentials page (once account is approved and whitelisted by Strong Compute)
Users can now choose between turning on isolated or shared workstations
Added new ENV variables accessible to running job commands:
NNODES
the total number of gpus in the clusterN_PROC
the number of gpus that the particular machine should use
Removed
nnodes
from isc file, replaced bygpus
field
UI/DX Improvements
We've added total training time to the experiment page
Burst billing calculates based on number of GPUs used
Bumped 10 dependencies
Bumped Elixir/Erlang ver.
Minor scheduler fixes
Burst self-cleans up tar files after use
Last updated