Restore future tokens into action expert module#246
Merged
youliangtan merged 7 commits intomainfrom Jul 9, 2025
Merged
Conversation
Signed-off-by: youliangt <youliangt@nvidia.com>
youliangtan
changed the title
HuFY-dev
approved these changes
Jul 8, 2025
Collaborator
HuFY-dev
left a comment
HuFY-dev
left a comment
There was a problem hiding this comment.
Fixed some minor issues. LGTM
Signed-off-by: youliangt <youliangt@nvidia.com>
|
Hi, what do u mean missing trained |
ddebenedittis
pushed a commit
to Borg-Robotics/Isaac-GR00T
that referenced
this pull request
Oct 7, 2025
* Restore future tokens into action experts module Signed-off-by: youliangt <youliangt@nvidia.com> * fix nit Signed-off-by: youliangt <youliangt@nvidia.com> * remove hack to load tensor Signed-off-by: youliangt <youliangt@nvidia.com> * nit Signed-off-by: youliangt <youliangt@nvidia.com> * Minor fixes * Format * update doc on top bottom image padding Signed-off-by: youliangt <youliangt@nvidia.com> --------- Signed-off-by: youliangt <youliangt@nvidia.com> Co-authored-by: HuFY-dev <127644049+HuFY-dev@users.noreply.github.com>
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR addresses the issue where the pretrained checkpoint being less effective in zeroshot scenario. This is mainly due to the missing trained
future_tokensembedding in the action expert during pretraining phase. 🐛action_head.future_tokens.weight. This would requires the latest updated checkpointUpdated model checkpoint: https://huggingface.co/nvidia/GR00T-N1.5-3B/discussions/5
Also some miscs update:
p/s: this is a breaking change for previously post-trained checkpoints.