Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
166 Open 2,663 Closed
166 Open 2,663 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

[Megatron-FSDP] Fix segmentation fault when using precision-aware optimizers with the updated MCore.
#3510 opened Apr 24, 2026 by cspades Contributor Loading…
5 tasks
@cspades
[docs] feat: document min GPU count for combined dense + MoE parallelism
#3509 opened Apr 24, 2026 by cuichenx Contributor Loading…
2 tasks
[ci] chore: bump .dev.commit to latest mcore-dev tested SHA docs-only With great power comes great responsibility.
#3507 opened Apr 23, 2026 by cuichenx Contributor Loading…
1 task
[vulnops][ckpt] fix: Use weights_only=True in TrainState checkpoint loading full-test-suite
#3506 opened Apr 23, 2026 by yaoyu-33 Contributor Loading…
3 of 4 tasks
2
Restore MegatronMIMO dataloader state when resuming training
#3503 opened Apr 23, 2026 by liding-nv Contributor • Draft
5
[ci, recipe] test: add L1 functional coverage for qwen3_vl forward step needs-more-tests Requires additional L0 and L1 test coverage before merge
#3502 opened Apr 23, 2026 by cuichenx Contributor Loading…
1 task
2
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-04-23) area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure work full-test-suite
#3496 opened Apr 23, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
chore(beep boop 🤖): Bump uv.lock (r0.4.0, mcore-core_r0.17.0) (2026-04-23) area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure work full-test-suite
#3495 opened Apr 23, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
auto-review PR's- Claude area:misc Cross-cutting utilities, logging, helpers, and other changes ci CI, automation, test queue, or workflow infrastructure work needs-review PR is ready for code review and waiting on a reviewer
#3494 opened Apr 23, 2026 by malay-nagda Contributor Loading…
5 tasks
3
cp: no recompute default (3470) into r0.4.0 area:perf Performance optimizations and benchmarking bug Something isn't working cherry-pick needs-review PR is ready for code review and waiting on a reviewer Run CICD
#3491 opened Apr 23, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
Use HybridEP flex dispatcher for Qwen3 235B B300 perf configs area:perf Performance optimizations and benchmarking feature New capabilities, enhancements, or enablement work
#3490 opened Apr 23, 2026 by rhmukundan Contributor • Draft
1
[training] fix: guard TECudaGraphHelper.delete_cuda_graphs() with graphs_created() area:training Training loop, callbacks, and runtime integration bug Something isn't working ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3488 opened Apr 22, 2026 by yaoyu-33 Contributor Loading…
2 tasks
4
[data] fix: raise on missing {% generation %} in chat template area:data Dataset builders, preprocessing, and samplers bug Something isn't working full-test-suite needs-review PR is ready for code review and waiting on a reviewer
#3486 opened Apr 22, 2026 by yaoyu-33 Contributor Loading…
4 tasks
1
3
[ci] fix: re-enable FSDP checkpoint test after TE 2.14 fix area:training Training loop, callbacks, and runtime integration ci CI, automation, test queue, or workflow infrastructure work
#3485 opened Apr 22, 2026 by yaoyu-33 Contributor Loading…
1 task
1
3
fix: pass total_tokens for SSM seq_idx in packed sequences area:training Training loop, callbacks, and runtime integration bug Something isn't working ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3484 opened Apr 22, 2026 by yaoyu-33 Contributor Loading…
3 tasks
1
3
[vulnops][data] fix: Validate URLs in VLM video loader to prevent SSRF 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
#3482 opened Apr 22, 2026 by yaoyu-33 Contributor Loading…
11 tasks done
2
[recipe] feat: enable THD packing by default for Qwen3.5-VL finetune area:recipe Training recipes and launch configs feature New capabilities, enhancements, or enablement work
#3481 opened Apr 22, 2026 by cuichenx Contributor • Draft
5 tasks
1
cp: Cleanup TE cuda graphs with the right api (3459) into r0.4.0 area:training Training loop, callbacks, and runtime integration bug Something isn't working cherry-pick needs-review PR is ready for code review and waiting on a reviewer Run CICD
#3476 opened Apr 22, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
Explicit error when the number of training samples < global batch size area:data Dataset builders, preprocessing, and samplers community-request feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer
#3464 opened Apr 22, 2026 by OlegSudakov Contributor Loading…
5 tasks
1
[model] fix: pass CPU initialization flag 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 needs-review PR is ready for code review and waiting on a reviewer ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3462 opened Apr 22, 2026 by pavelgein Contributor Loading…
1 of 4 tasks
3
[Bugfix] MiniMax-M2 full-dimension k_norm sharding when tp_size > num_kv_heads area:model Model implementations and HF bridge logic bug Something isn't working community-request needs-review PR is ready for code review and waiting on a reviewer
#3458 opened Apr 21, 2026 by HollowMan6 Contributor Loading…
2 of 5 tasks
12
[examples] Rename diffusion/recipes to diffusion/models area:diffusion DFM module feature New capabilities, enhancements, or enablement work needs-author Author action is required before review or merge can continue ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3455 opened Apr 21, 2026 by cuichenx Contributor Loading…
4
[ckpt] fix: use MSC paths area:ckpt Checkpoint conversion, loading, export, and save paths bug Something isn't working community-request needs-review PR is ready for code review and waiting on a reviewer ready-to-merge PR is approved, current, and only waiting for CI to pass before merge waiting-on-customer Waiting on the original author to respond
#3452 opened Apr 21, 2026 by pavelgein Contributor Loading…
1 of 5 tasks
3
upstream MCore tokenizers config area:training Training loop, callbacks, and runtime integration feature New capabilities, enhancements, or enablement work
#3451 opened Apr 21, 2026 by dimapihtar Contributor • Draft
5 tasks
1
[PEFT] fix: fused fc1 projection mapping for MiniMax (w1 & w3) area:peft Parameter-efficient fine-tuning (LoRA, adapters) bug Something isn't working community-request needs-review PR is ready for code review and waiting on a reviewer ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3449 opened Apr 21, 2026 by HollowMan6 Contributor Loading…
2 of 5 tasks
6
ProTip! Follow long discussions with comments:>50.