[Feature]: add ViT activation_offload for InternS1#1619
Open
NengXu001 wants to merge 1 commit intoInternLM:mainfrom
Open
[Feature]: add ViT activation_offload for InternS1#1619NengXu001 wants to merge 1 commit intoInternLM:mainfrom
NengXu001 wants to merge 1 commit intoInternLM:mainfrom
Conversation
HAOCHENYE
reviewed
Mar 25, 2026
Collaborator
HAOCHENYE
left a comment
HAOCHENYE
left a comment
There was a problem hiding this comment.
We should enhance the implementation of activation offload to make it more general-purpose, so that it can serve multiple models without needing to be aware of layer counts.
| use_mask_token: bool = False | ||
| use_mean_pooling: bool = True | ||
| attn_impl: Literal["flash_attention", "flex_attention", "eager_attention"] = "flash_attention" | ||
| text_hidden_layers: int = 0 |
Collaborator
There was a problem hiding this comment.
We should not modify other modules to work around functional deficiencies in ActivationOffload itself. I understand that, from VisionConfig's perspective, it should not need to be aware of text_hidden_layers.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Reduce InternS1 training memory by ~10GB via ViT offloading: