Megatron-DeepSpeed
Compute model param count once
#204
Open
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Overview
Commits
6
Changes
View On
GitHub
Compute model param count once
#204
jaketae
wants to merge 6 commits into
main
from
rm-duplicate-param-count
refactor: compute model param count once
544108dc
jaketae
marked this pull request as ready for review
4 years ago
jaketae
requested a review
from
stas00
4 years ago
stas00
commented on 2021-11-24
stas00
requested a review
from
TevenLeScao
4 years ago
Update megatron/training.py
816b8670
jaketae
commented on 2021-11-25
Update megatron/training.py
c2d63903
fix: use deepspeed param count method
f4c7c67e
refactor: replace filter w/ list comp, generator to list
a7b10b7c
refactor: use set for constant time lookup
ac3e138b
Login to write a write a comment.
Login via GitHub
Reviewers
stas00
TevenLeScao
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub