DeepSpeed
[Engine] Only scale gradients if scale_wrt_gas is True
#7724
Merged

Commits
  • Only scale gradients if scale_wrt_gas is True
    kashif committed 39 days ago
  • add test to verify scale_wrt_gas=False
    tohtana committed 39 days ago
  • Merge pull request #1 from tohtana/tohtana/add_test_disable_scaling
    kashif committed 38 days ago
  • Merge branch 'master' into fix-hook
    kashif committed 38 days ago
  • dry up the method
    kashif committed 38 days ago
Loading