Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

force fp4_param to false
#5307 opened Aug 4, 2026 by nv-mollys Contributor Draft
5 tasks
fix(model): make gemma4_vl inference run end to end 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
#5305 opened Aug 4, 2026 by pruprakash Contributor Loading…
feat(nemotron-omni): enable temporal video training area:model Model implementations and HF bridge logic feature New capabilities, enhancements, or enablement work full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge waiting-on-customer Waiting on the original author to respond
#5304 opened Aug 4, 2026 by ehosseiniasl Loading…
[recipes] refactor: standardize Nemotron 3.5 Lightning naming area:recipe Training recipes and launch configs breaking-change Public behavior or API compatibility changes feature New capabilities, enhancements, or enablement work 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
#5303 opened Aug 4, 2026 by cuichenx Contributor Loading…
fix(ckpt): validate conversion world size against TP*PP, not TP*PP*EP 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
#5302 opened Aug 4, 2026 by pruprakash Contributor Loading…
cp: feat(model): add GLM-5.2 verification card (5034) into r0.6.0 area:recipe Training recipes and launch configs cherry-pick community-request feature New capabilities, enhancements, or enablement work full-test-suite label-card Model verification card 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 r0.6.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge. Run CICD
#5301 opened Aug 4, 2026 by svcnvidia-nemo-ci Contributor Loading…
cp: Nemotron 3.5 Lightning checkpoint and naming updates (#5293, #5303) into r0.6.0 area:recipe Training recipes and launch configs breaking-change Public behavior or API compatibility changes bug Something isn't working cherry-pick full-test-suite label-card Model verification card 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 Run CICD
#5300 opened Aug 4, 2026 by svcnvidia-nemo-ci Contributor Loading…
fix(perf): lock memory clock per GPU
#5299 opened Aug 4, 2026 by malay-nagda Contributor Draft
fix(quant): wrap post-quantization VLM generation in torch.no_grad() area:quant Quantization (PTQ, QAT, FP8 recipes) 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
#5292 opened Aug 4, 2026 by Batool1227 Loading…
3 tasks done
feat(data): blend packed Parquet SFT sources
#5289 opened Aug 4, 2026 by yaoyu-33 Contributor Draft
cp: fix(training): pad uneven HybridEP dispatch inputs (5270) into r0.6.0 area:training Training loop, callbacks, and runtime integration bug Something isn't working cherry-pick 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 r0.6.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge. Run CICD
#5288 opened Aug 4, 2026 by cuichenx Contributor Loading…
feat (recipe): add Nemotron 3 Ultra MXFP8 GB200 pretraining perf-recipe 26.08 area:perf Performance optimizations and benchmarking feature New capabilities, enhancements, or enablement work full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge performance/release Performance items related with NeMo release performance 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
#5287 opened Aug 4, 2026 by ghadiaravi13 Contributor Loading…
5 tasks
26.08
build(deps): upgrade Transformers to 5.14.1 area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure work 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
#5286 opened Aug 3, 2026 by bvolpato Loading…
4 tasks done
docs(models): record Gemma 4 SFT verification blockers 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
#5283 opened Aug 3, 2026 by yaoyu-33 Contributor Loading…
Add MLPerf Llama 3.1 8B Perf Recipe
#5277 opened Aug 3, 2026 by nv-mollys Contributor Draft
5 tasks
docs(ckpt): document Megatron-LM checkpoint export area:ckpt Checkpoint conversion, loading, export, and save paths docs Documentation-only updates or documentation debt docs-only With great power comes great responsibility.
#5276 opened Aug 3, 2026 by yaoyu-33 Contributor Draft
5 tasks done
add nemotron puzzle support area:model Model implementations and HF bridge logic blocked Work cannot move forward until an external dependency is cleared feature New capabilities, enhancements, or enablement work full-test-suite needs-more-tests Requires additional L0 and L1 test coverage before merge
#5265 opened Aug 3, 2026 by rohitrango Contributor Loading…
1 of 5 tasks
fix(ckpt): skip unmapped conversion task slots instead of crashing area:model Model implementations and HF bridge logic 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
#5262 opened Aug 3, 2026 by bzantium Contributor Loading…
2 of 5 tasks
perf(recipe): add DeepSeek V4 Flash GB300 FP8MX recipes 26.08 area:perf Performance optimizations and benchmarking 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 r0.6.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge.
#5258 opened Aug 3, 2026 by malay-nagda Contributor Loading…
4 tasks done
sjayaram/k8s excalibur community-request needs-review PR is ready for code review and waiting on a reviewer
#5254 opened Aug 3, 2026 by sjayaram-nv Draft
fix(qwen3.5-vl): use top-level prefix for dense MTP mappings 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
#5253 opened Aug 2, 2026 by worktrial2-trajectory Loading…
chore(deps): AUT-1290 bump MCore main and validate dev area:build Dependencies, packaging, images, and environment setup ci CI, automation, test queue, or workflow infrastructure 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
#5250 opened Aug 2, 2026 by svcnemo-autobot Contributor Loading…
test(deps): AUT-1290 validate Bridge against MCore dev area:ci ci CI, automation, test queue, or workflow infrastructure work dummy-pr Filed this PR to run tests, not going to merge full-test-suite
#5251 opened Aug 2, 2026 by svcnemo-autobot Contributor Loading…
fix(diffusion): calculate WAN FLOPs from runtime geometry area:diffusion DFM module area:perf Performance optimizations and benchmarking 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
#5249 opened Aug 2, 2026 by cuichenx Contributor Loading…
4 tasks done
ProTip! Follow long discussions with comments:>50.