Skip to content
This repository has been archived by the owner on Oct 19, 2024. It is now read-only.

[FEATURE] Current gradient accumulation only works for jnp.mean loss #268

Open
zhuohan123 opened this issue Dec 16, 2021 · 0 comments
Open

[FEATURE] Current gradient accumulation only works for jnp.mean loss #268

zhuohan123 opened this issue Dec 16, 2021 · 0 comments
Labels
enhancement New feature

Comments

@zhuohan123
Copy link
Member

zhuohan123 commented Dec 16, 2021 •
edited by merrymercy
Loading

The current gradient accumulation only works for jnp.mean loss because we always use mean reduction.
For other losses or auxiliary states, we should support other reduction types such as sum reduction and concatenation reduction.
@zhuohan123 zhuohan123 added the enhancement New feature label Dec 26, 2021
@merrymercy merrymercy changed the title Current gradient accumulation only works for jnp.mean loss [FEATURE] Current gradient accumulation only works for jnp.mean loss Mar 11, 2022
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
enhancement New feature
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@zhuohan123