Merged
Conversation
Signed-off-by: jinliangl <jinliangl@nvidia.com>
Signed-off-by: jinliangl <jinliangl@nvidia.com>
…rgency has not been test yet.
…ck's param, need to check it further.
Signed-off-by: jinliangl <jinliangl@nvidia.com>
…ision_model=true to enable it Signed-off-by: jinliangl <jinliangl@nvidia.com>
Signed-off-by: jinliangl <jinliangl@nvidia.com>
Signed-off-by: jinliangl <jinliangl@nvidia.com>
Signed-off-by: jinliangl <jinliangl@nvidia.com>
Signed-off-by: Shifang Xu <shifangx@nvidia.com>
shifangx
added a commit
that referenced
this pull request
Jan 29, 2026
* qwen3-vl migration [wip] Signed-off-by: jinliangl <jinliangl@nvidia.com> * support qwen3vl bshd training Signed-off-by: jinliangl <jinliangl@nvidia.com> * support bshd training, thd training, thd training with cp, loss convergency has not been test yet. * adjust how to get packed seq params * supoort qwen3vl dense * support convert hf ckpt for qwenvl vision module, can convert deepstack's param, need to check it further. * fix qwen3vl shard_state_dict * align megagtron-bridge and mbridge fwd bitwisely * align bshd and thd training, cp training remains to do * fix full recompute of qwen3vl bug with credit to xuwen Signed-off-by: jinliangl <jinliangl@nvidia.com> * support hf vision model and megatron vision model; use model.use_hf_vision_model=true to enable it Signed-off-by: jinliangl <jinliangl@nvidia.com> * align with pr 1997, thd and bshd loss curve is verified Signed-off-by: jinliangl <jinliangl@nvidia.com> * init hf_config from scratch to avoid deepcopy parallel_state Signed-off-by: jinliangl <jinliangl@nvidia.com> * use pg_collection instead of mpu Signed-off-by: jinliangl <jinliangl@nvidia.com> * add qwen3vl_step new file --------- Signed-off-by: jinliangl <jinliangl@nvidia.com> Signed-off-by: Shifang Xu <shifangx@nvidia.com> Co-authored-by: jinliangl <jinliangl@nvidia.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
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.
What does this PR do ?
Add a one line overview of what this PR aims to accomplish.
Changelog
GitHub Actions CI
See the CI sectionin the Contributing doc for how to trigger the CI. A Nvidia developer will need to approve and trigger the CI for external contributors.
Before your PR is "Ready for review"
Pre checks:
If you haven't finished some of the above items you can still open "Draft" PR.
Additional Information