sparseml
sparseml copied to clipboard
don't save epoch on IC one_shot checkpoints
default saved epoch for one_shot in the IC flows is -1 due to Trainer initialization. This will cause issues on model load since the checkpoint recipe will be initialized to epoch -1, not applying any of the optimizations. This PR removes saving the epoch for one shot models so the entire one shot recipe will be applied on checkpoint load.
test_plan: @anmarques to verify
@anmarques @rahul-tuli assigned for review