Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
166 Open 2,498 Closed
166 Open 2,498 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

[training] fix: use int64 for TrainState counters to prevent overflow
#3312 opened Apr 14, 2026 by yaoyu-33 Contributor Loading…
1 of 2 tasks
3
perf(nsys): reduce CPU-side overhead in profiling defaults
#3311 opened Apr 13, 2026 by dingqingy-nv Contributor • Draft
3 tasks
2
cp: fix(perf): set NCCL env vars when nccl_ub enabled via recipe config (3283) into r0.4.0
#3305 opened Apr 13, 2026 by yaoyu-33 Contributor • Draft
2 tasks
6
[ckpt] feat: support MSC for fsdp_dtensors area:ckpt Checkpoint conversion, loading, export, and save paths community-request ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#3300 opened Apr 13, 2026 by pavelgein Contributor Loading…
3 of 5 tasks
1
5
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-04-13) area:build Dependencies, packaging, images, and environment setup full-test-suite needs-review PR is ready for code review and waiting on a reviewer
#3297 opened Apr 13, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
[doc] feat: add MiniMax M2.5 / M2.7 model support docs-only With great power comes great responsibility. needs-review PR is ready for code review and waiting on a reviewer
#3291 opened Apr 13, 2026 by yaoyu-33 Contributor Loading…
4 tasks done
2
[ckpt] fix: Prevent int32 overflow for high train sample counts. area:ckpt Checkpoint conversion, loading, export, and save paths bug Something isn't working community-request needs-author Author action is required before review or merge can continue
#3290 opened Apr 12, 2026 by BlueCrescent • Draft
3
[model] feat: Add YARN support for mamba_model from MCORE area:model Model implementations and HF bridge logic feature New capabilities, enhancements, or enablement work needs-author Author action is required before review or merge can continue
#3289 opened Apr 12, 2026 by guihong-nv • Draft
3 of 5 tasks
1
feat(perf): switch llama3 70B GB200/GB300 MXFP8 from 3D parallel to FSDP area:perf Performance optimizations and benchmarking needs-author Author action is required before review or merge can continue performance
#3284 opened Apr 12, 2026 by dingqingy-nv Contributor • Draft
1
[test] refactor: move diffusion tests to test_groups directory area:diffusion DFM module r0.4.0 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
#3275 opened Apr 10, 2026 by huvunvidia Contributor Loading…
2 tasks
3
docs: expand golden values update strategy in release process guide docs-only With great power comes great responsibility. needs-review PR is ready for code review and waiting on a reviewer
#3268 opened Apr 10, 2026 by ko3n1g Contributor Loading…
2 tasks
4
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-04-10) area:build Dependencies, packaging, images, and environment setup full-test-suite needs-review PR is ready for code review and waiting on a reviewer
#3264 opened Apr 10, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
[bridge] feat: Add ERNIE 4.5 text-only MoE and VL bridges for Megatro… area:model Model implementations and HF bridge logic community-request
#3263 opened Apr 10, 2026 by bo-ke Loading…
17
cp: [perf] fix: guard cuda_graph_scope validation against None (3249) into r0.4.0 area:perf Performance optimizations and benchmarking cherry-pick needs-author Author action is required before review or merge can continue Run CICD
#3262 opened Apr 10, 2026 by svcnvidia-nemo-ci Contributor • Draft
6
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-04-09) full-test-suite
#3242 opened Apr 9, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
Comment out dfm imports area:recipe Training recipes and launch configs bug Something isn't working needs-author Author action is required before review or merge can continue needs-follow-up Issue needs follow-up needs-review PR is ready for code review and waiting on a reviewer
#3233 opened Apr 9, 2026 by gautham-kollu Contributor Loading…
5 tasks
7
Adding condition for how nccl_chunk_size var is being set as currentl… area:perf Performance optimizations and benchmarking bug Something isn't working docs-only With great power comes great responsibility. 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
#3228 opened Apr 8, 2026 by rsalagame-nvidia Contributor Loading…
@rsalagame-nvidia
4
Adding condition for how nccl_chunk_size var is being set as currentl… area:perf Performance optimizations and benchmarking bug Something isn't working needs-author Author action is required before review or merge can continue
#3226 opened Apr 8, 2026 by rsalagame-nvidia Contributor Loading…
@rsalagame-nvidia
5
Expose fully_parallel_save in save_megatron_model area:ckpt Checkpoint conversion, loading, export, and save paths community-request feature New capabilities, enhancements, or enablement work needs-author Author action is required before review or merge can continue needs-follow-up Issue needs follow-up 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
#3207 opened Apr 8, 2026 by nic-nvidia Loading…
1 of 2 tasks
4
[misc] feat: Add NVTX ranges for train_step and optimizer_step area:misc Cross-cutting utilities, logging, helpers, and other changes feature New capabilities, enhancements, or enablement work
#3198 opened Apr 7, 2026 by minitu • Draft
5 tasks
1
fix(mimo): Scale encoder gradients for heterogeneous DP in multimodule finalization area:training Training loop, callbacks, and runtime integration bug Something isn't working needs-review PR is ready for code review and waiting on a reviewer
#3197 opened Apr 7, 2026 by kamran-nvidia Contributor Loading…
5 tasks
1
Enable nemo-ci tests (short runs - perf and non-perf) for Wan + Updating recipes names area:diffusion DFM module area:recipe Training recipes and launch configs ci CI, automation, test queue, or workflow infrastructure work
#3179 opened Apr 6, 2026 by huvunvidia Contributor Loading…
5 tasks
17
ci: Add FLUX/diffusion support to scripts/performance/run_recipe.py area:diffusion DFM module ci CI, automation, test queue, or workflow infrastructure work needs-review PR is ready for code review and waiting on a reviewer
#3176 opened Apr 6, 2026 by suiyoubi Contributor Loading…
5 tasks
7
Update Qwen3 235B B300 to include VP area:perf Performance optimizations and benchmarking
#3175 opened Apr 6, 2026 by rhmukundan Contributor • Draft
1
[model] feat: add MiMo-V2-Flash model support community-request
#3163 opened Apr 5, 2026 by beccohov • Draft
3 of 5 tasks
7
ProTip! Adding no:label will show everything without a label.