remove legacy tranformer and modules by dimapihtar · Pull Request #4207 · NVIDIA/Megatron-LM

dimapihtar · 2026-04-08T13:55:39Z

What does this PR do ?

Removes legacy tranformer and other legacy modules as a part of MLM code cleanup.

⚠️ For major changes (either in lines of code or in its impact), please make sure to first share a design doc with the team. If you're unsure what's the best way to do so, contact the @mcore-oncall.

Contribution process

Pre-checks

I have added relevant unit tests
I have added relevant functional tests
I have added proper typing to my code Typing guidelines
I have added relevant documentation
I have run the autoformatter.sh on my PR

Code review

Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!

All PRs start as draft. If you open a non-draft PR, it will be automatically converted to draft.

Step 1: Mark PR as "Ready for Review"

When your PR is ready, click Ready for Review.
An oncall reviewer is auto-assigned and expert reviewers are notified based on your changes.
- Some PRs may jump straight to step 2. This is determined by .github/CODEOWNERS.

⚠️ Only mark as ready once merge-conflicts are resolved and the CI is passing.
Final Review might get declined if these requirements are not fulfilled.

Step 2: Final Review

For PRs that change megatron/core, once all expert reviewers have approved, the Final Review label is applied automatically and final reviewers are assigned.

For PRs outside megatron/core, this step is skipped.

Step 3: Approved

Once all required reviewers have approved, the Approved label is applied automatically.

Merge

Any member of mcore-engineers will be able to merge your PR.

For MRs into `dev` branch

The proposed review process for `dev` branch is under active discussion.

MRs are mergable after one approval by either eharper@nvidia.com or zijiey@nvidia.com.

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

copy-pr-bot · 2026-04-08T13:55:43Z

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

dimapihtar · 2026-04-08T13:56:04Z

/ok to test 53b28f3

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

dimapihtar · 2026-04-08T15:00:43Z

/ok to test 9dc4c7b

lmcafee-nvidia · 2026-04-21T15:40:23Z

all of the model files (gpt_model.py, bert_model.py, t5_model.py, etc.) still point to this file. have these been removed in other PRs, or do they need to be removed in this PR?

yes, bert & t5 models were already removed

#4203
#4204

gpt model PR is ongoing and will be merged as well: #4322

lmcafee-nvidia · 2026-04-21T15:41:49Z

removing LayerNorm and RMSNorm also breaks existing model files. We need to either remove the models, or merge in main if they've already been removed.

lmcafee-nvidia · 2026-04-21T15:44:54Z

a few things that still depend on this module:

megatron/training/checkpointing.py

tools/checkpoint/loader_legacy.py

examples/academic_paper_scripts/detxoify_lm/generate_samples_gpt.py

examples/academic_paper_scripts/detxoify_lm/finetune_gpt.py

examples/gpt3/gpt_config.yaml

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

dimapihtar · 2026-05-05T22:25:20Z

/ok to test 56ba238

svcnvidia-nemo-ci · 2026-05-05T23:56:12Z

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/25408874973

svcnvidia-nemo-ci · 2026-05-06T00:39:00Z

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/25410186567

dimapihtar · 2026-05-06T12:36:39Z

/ok to test 0d4fc55

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

dimapihtar · 2026-05-06T18:36:26Z

/ok to test a1a6b5d

svcnvidia-nemo-ci · 2026-05-07T11:59:52Z

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/25494423909

…ync-2026-05 Upstream tip: 932d9ee (Add periodic GPU sniff tests, 2026-05-07). Conflict resolution: - 13 modify/delete conflicts in .github/* and skills/*: kept deletion (matches policy from c037427 removing upstream-NVIDIA CI machinery not relevant to this fork). - AGENTS.md content conflict: kept fork version (skills/contributing sections from upstream don't apply since the directories were removed). - New upstream additions matching the same removal policy were dropped: .claude/settings.json, .github/workflows/nightly-sync-main-to-dev.yml, skills/{cicd,linting-and-formatting,nightly-sync,run-on-slurm,testing, update-golden-values}/SKILL.md. - SECURITY.md (new from upstream): kept (generic security policy). Verified post-merge: - pretrain_gpt.py logging-patch hook intact (lines 405-408). - megatron/core/optimizer/__init__.py emerging_optimizers + muon logic intact. - megatron/training/arguments.py custom muon flags intact (lines 2297-2309). Notable upstream changes pulled in: - Removed legacy transformer + legacy GPT modules (NVIDIA#4207, NVIDIA#4322). - Docker base image bump to 26.04-py3 (NVIDIA#4611). - Inference fixes (vLLM grouped GEMM NVIDIA#4566, FlashInfer sampling NVIDIA#2456, EP sync NVIDIA#4607, MoE dispatcher fixes NVIDIA#4576). - Gradient corruption fix with layerwise param all-gather overlap (NVIDIA#4609). - Hybrid model + Flextron + GPU sniff tests + named validation sets + fault injection + InJob restart.

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: yhgalaxy <yhgalaxy@outlook.com>

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: Jon Barker <jbarker@aws-cmh-slurm-1-vscode-02.cm.cluster>

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: Dmytro Pykhtar <dpykhtar@nvidia.com>

remove legacy tranformer and modules

53b28f3

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

dimapihtar added complexity: low Expert Review [deprecated] Apply this label to indicate that your PR is ready for expert review. labels Apr 8, 2026

svcnvidia-nemo-ci added this to the Core 0.16 milestone Apr 8, 2026

copy-pr-bot Bot temporarily deployed to test April 8, 2026 13:57 Inactive

dimapihtar marked this pull request as ready for review April 8, 2026 14:52

svcnvidia-nemo-ci requested a review from a team April 8, 2026 14:52

svcnvidia-nemo-ci added complexity: high and removed complexity: low labels Apr 8, 2026

remove legacy imports

9dc4c7b

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

copy-pr-bot Bot temporarily deployed to test April 8, 2026 15:02 Inactive

dimapihtar requested review from deepakn94 and lmcafee-nvidia April 8, 2026 16:18

lmcafee-nvidia reviewed Apr 21, 2026

View reviewed changes

dimapihtar added 2 commits April 27, 2026 09:13

Merge branch 'main' into remove_legacy_transformer

9c5e2f0

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

fix docs

445e107

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

dimapihtar requested a review from jaredcasper May 5, 2026 22:21

jaredcasper approved these changes May 5, 2026

View reviewed changes

Merge branch 'main' into remove_legacy_transformer

56ba238

copy-pr-bot Bot temporarily deployed to test May 5, 2026 22:26 Inactive

dimapihtar added this pull request to the merge queue May 5, 2026

github-merge-queue Bot removed this pull request from the merge queue due to failed status checks May 6, 2026

dimapihtar added this pull request to the merge queue May 6, 2026

dimapihtar removed this pull request from the merge queue due to a manual request May 6, 2026

dimapihtar added this pull request to the merge queue May 6, 2026

dimapihtar removed this pull request from the merge queue due to a manual request May 6, 2026

Merge branch 'main' into remove_legacy_transformer

0d4fc55

copy-pr-bot Bot temporarily deployed to test May 6, 2026 12:37 Inactive

Merge branch 'NVIDIA:main' into main

106a0be

dimapihtar added this pull request to the merge queue May 6, 2026

github-merge-queue Bot removed this pull request from the merge queue due to a conflict with the base branch May 6, 2026

dimapihtar and others added 2 commits May 6, 2026 21:31

Merge branch 'NVIDIA:main' into main

422590f

Merge branch 'main' into remove_legacy_transformer

a1a6b5d

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

copy-pr-bot Bot temporarily deployed to test May 6, 2026 18:37 Inactive

dimapihtar enabled auto-merge May 7, 2026 11:30

dimapihtar added this pull request to the merge queue May 7, 2026

Merged via the queue into NVIDIA:main with commit 8de8238 May 7, 2026
71 of 73 checks passed

dimapihtar deleted the remove_legacy_transformer branch May 7, 2026 12:33

This was referenced May 7, 2026

chore: nightly sync main into dev (07_05_2026) #4684

Closed

chore: nightly sync main into dev (08_05_2026) #4708

Closed

chore: nightly sync main into dev (11_05_2026) #4739

Closed

svcnvidia-nemo-ci mentioned this pull request May 14, 2026

chore: nightly sync main into dev (14_05_2026) #4809

Closed

yhgalaxy pushed a commit to yhgalaxy/Megatron-LM that referenced this pull request Jun 17, 2026

remove legacy tranformer and modules (NVIDIA#4207)

0da4642

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: yhgalaxy <yhgalaxy@outlook.com>

jon-barker pushed a commit to jon-barker/Megatron-LM that referenced this pull request Jul 10, 2026

remove legacy tranformer and modules (NVIDIA#4207)

fecc7fd

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: Jon Barker <jbarker@aws-cmh-slurm-1-vscode-02.cm.cluster>

terminator123 pushed a commit to 021ai/Megatron-LM that referenced this pull request Aug 3, 2026

remove legacy tranformer and modules (NVIDIA#4207)

55047ea

Signed-off-by: dimapihtar <dpykhtar@nvidia.com>

svcnvidia-nemo-ci pushed a commit to dimapihtar/Megatron-LM that referenced this pull request Aug 4, 2026

remove legacy tranformer and modules (NVIDIA#4207)

125eed6

Signed-off-by: dimapihtar <dpykhtar@nvidia.com> Signed-off-by: Dmytro Pykhtar <dpykhtar@nvidia.com>

Uh oh!

Conversation

dimapihtar commented Apr 8, 2026

What does this PR do ?

Contribution process

Pre-checks

Code review

Step 1: Mark PR as "Ready for Review"

Step 2: Final Review

Step 3: Approved

Merge

Uh oh!

copy-pr-bot Bot commented Apr 8, 2026

Uh oh!

dimapihtar commented Apr 8, 2026

Uh oh!

dimapihtar commented Apr 8, 2026

Uh oh!

lmcafee-nvidia Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

dimapihtar Apr 27, 2026

Choose a reason for hiding this comment

Uh oh!

lmcafee-nvidia Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

lmcafee-nvidia Apr 21, 2026

Choose a reason for hiding this comment

Uh oh!

dimapihtar commented May 5, 2026

Uh oh!

svcnvidia-nemo-ci commented May 5, 2026

Uh oh!

svcnvidia-nemo-ci commented May 6, 2026

Uh oh!

Uh oh!

dimapihtar commented May 6, 2026

Uh oh!

dimapihtar commented May 6, 2026

Uh oh!

svcnvidia-nemo-ci commented May 7, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants