DeepSpeed
Support fp32 grad clipping and fix max_grad_norm confusion
#232
Merged
Overview
Commits
8
Changes
View On GitHub

Support fp32 grad clipping and fix max_grad_norm confusion #232

jeffra merged 8 commits into master from jeffra/max_grad_update
jeffra
jeffra updates to support fp32 grad clipping and disable max_grad_norm support
c44360b5
jeffra jeffra requested a review from tjruwase tjruwase 5 years ago
jeffra revert unused function
58a7dbc6
jeffra clipping default 0
77ea355c
jeffra jeffra requested a review from samyam samyam 5 years ago
tjruwase
tjruwase approved these changes on 2020-05-26
jeffra update tests
aa59de69
jeffra update tests and fix bug
ee893de5
jeffra Merge branch 'master' into jeffra/max_grad_update
b4422b97
jeffra rename inside fp32 test
811c986d
jeffra Merge branch 'jeffra/max_grad_update' of github.com:microsoft/DeepSpe…
8dd7acdd
samyam
samyam commented on 2020-05-27
samyam
samyam approved these changes on 2020-05-27
jeffra jeffra merged abe2204d into master 5 years ago
jeffra jeffra deleted the jeffra/max_grad_update branch 5 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum