Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
139 Open 4,594 Closed
139 Open 4,594 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[misc] fix: Cap server requests by token budget area:misc Cross-cutting utilities, logging, helpers, and other changes bug Something isn't working needs-review PR is ready for code review and waiting on a reviewer
#5594 opened Aug 15, 2026 by yaoyu-33 Contributor Loading…
4
[ckpt] fix: Restore checkpoints without Energon state area:ckpt Checkpoint conversion, loading, export, and save paths bug Something isn't working needs-review PR is ready for code review and waiting on a reviewer
#5593 opened Aug 15, 2026 by yaoyu-33 Contributor Loading…
3
fix(model): preserve masked placeholder embeddings area:model Model implementations and HF bridge logic bug Something isn't working needs-review PR is ready for code review and waiting on a reviewer
#5592 opened Aug 15, 2026 by yaoyu-33 Contributor Loading…
4
[training] fix: Synchronize MIMO loss scalers area:training Training loop, callbacks, and runtime integration bug Something isn't working needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5591 opened Aug 15, 2026 by yaoyu-33 Contributor Loading…
3
fix(model): use dynamic temporal video token counts
#5590 opened Aug 14, 2026 by cuichenx Contributor • Draft
4 tasks done
1
feat(optim): add LoRA+ learning-rate split for Megatron LoRA area:peft Parameter-efficient fine-tuning (LoRA, adapters) community-request feature New capabilities, enhancements, or enablement work needs-more-tests Requires additional L0 and L1 test coverage before merge ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5589 opened Aug 14, 2026 by HollowMan6 Contributor Loading…
3 of 5 tasks
3
feat(peft): expand_shared_outer export flag for shared-outer MoE LoRA area:peft Parameter-efficient fine-tuning (LoRA, adapters) community-request feature New capabilities, enhancements, or enablement work needs-more-tests Requires additional L0 and L1 test coverage before merge ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5588 opened Aug 14, 2026 by HollowMan6 Contributor Loading…
2 of 5 tasks
2
feat(qwen): verify Qwen 3.8 27B and tune GB200 pretraining area:model Model implementations and HF bridge logic bug Something isn't working needs-more-tests Requires additional L0 and L1 test coverage before merge ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5587 opened Aug 14, 2026 by cuichenx Contributor Loading…
5 tasks done
1
fix(peft): support hybrid-stack activation recompute community-request
#5586 opened Aug 14, 2026 by j316chuck • Draft
1
fix(ckpt): skip export of parameters no PP rank owns area:ckpt Checkpoint conversion, loading, export, and save paths bug Something isn't working full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer qa_rcca_done
#5585 opened Aug 14, 2026 by pruprakash Contributor Loading…
2 of 5 tasks
3
docs: add DeepSeek V3 verification card area:model Model implementations and HF bridge logic docs Documentation-only updates or documentation debt docs-only With great power comes great responsibility. needs-review PR is ready for code review and waiting on a reviewer
#5583 opened Aug 14, 2026 by yaoyu-33 Contributor Loading…
3 tasks done
3
feat(qwen): add Qwen3.5 sequence classification bridge area:model Model implementations and HF bridge logic community-request feature New capabilities, enhancements, or enablement work full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5580 opened Aug 14, 2026 by nidhishs Loading…
7 of 8 tasks
2
feat(training): wire generalized tensor parallelism area:training Training loop, callbacks, and runtime integration feature New capabilities, enhancements, or enablement work full-test-suite high-complexity Harder to merge: prone to conflicts and needs additional test coverage needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5569 opened Aug 13, 2026 by yaoyu-33 Contributor Loading…
3
feat(recipe): add Nemotron 3 Ultra verification recipes area:recipe Training recipes and launch configs feature New capabilities, enhancements, or enablement work high-complexity Harder to merge: prone to conflicts and needs additional test coverage model-nemotron
#5568 opened Aug 13, 2026 by cuichenx Contributor • Draft
3 of 5 tasks
1
fix: PEFT parameter statistics only count the first model chunk area:peft Parameter-efficient fine-tuning (LoRA, adapters) bug Something isn't working community-request ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5562 opened Aug 13, 2026 by andrewwhitecdw Loading…
6
fix: token_dtype_code assigns uint16 to vocab sizes needing uint64 area:data Dataset builders, preprocessing, and samplers bug Something isn't working community-request ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5561 opened Aug 13, 2026 by andrewwhitecdw Loading…
3
feat(recipes): modernize DeepSeek V4 library configs
#5554 opened Aug 13, 2026 by cuichenx Contributor • Draft
1 task done
1
fix(model): normalize Qwen3-Omni integer masks for TE area:model Model implementations and HF bridge logic bug Something isn't working community-request needs-more-tests Requires additional L0 and L1 test coverage before merge ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5544 opened Aug 13, 2026 by hbhflw2000 Contributor Loading…
3
[4/4] fix(data): align thinking-model chat templates and masks area:data Dataset builders, preprocessing, and samplers bug Something isn't working needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5541 opened Aug 12, 2026 by yaoyu-33 Contributor Loading…
4
feat(training): enable MFSDP V2 gradient clipping
#5540 opened Aug 12, 2026 by wujingyue Contributor • Draft
1
reuse save_tokenizer_assets from MLM [not ready for review]
#5532 opened Aug 12, 2026 by dimapihtar Contributor • Draft
5 tasks
1
feat(data): scalable canonical-grid reads for MegatronMIMO data loading area:data Dataset builders, preprocessing, and samplers community-request feature New capabilities, enhancements, or enablement work full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5525 opened Aug 12, 2026 by going-song Contributor Loading…
5
perf(recipe): tune Nemotron Omni H100 fine-tuning area:perf Performance optimizations and benchmarking feature New capabilities, enhancements, or enablement work full-test-suite high-complexity Harder to merge: prone to conflicts and needs additional test coverage needs-more-tests Requires additional L0 and L1 test coverage before merge ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5524 opened Aug 12, 2026 by cuichenx Contributor Loading…
4 tasks done
5
fix(quant): dequantize MXFP4 on GPU under the distributed conversion backend area:quant Quantization (PTQ, QAT, FP8 recipes) bug Something isn't working community-request full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge needs-review PR is ready for code review and waiting on a reviewer
#5523 opened Aug 12, 2026 by mikan-atomoki Loading…
1 of 5 tasks
2
fix(ckpt): pass device_ids to barrier in GPU conversion backend area:ckpt Checkpoint conversion, loading, export, and save paths bug Something isn't working community-request needs-more-tests Requires additional L0 and L1 test coverage before merge r0.6.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge. ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#5522 opened Aug 12, 2026 by mikan-atomoki Loading…
1 of 5 tasks
5
ProTip! no:milestone will show everything without a milestone.