Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
163 Open 3,421 Closed
163 Open 3,421 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

fix(deepseek): gate H100 fused kernel defaults area:perf Performance optimizations and benchmarking bug Something isn't working ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#4338 opened Jun 12, 2026 by cuichenx Contributor Loading…
2
cp: [recipe,model,ckpt] DeepSeek-V4 backports (4131, 4271, 4305, 4306, 4338) into r0.5.0 area:recipe Training recipes and launch configs cherry-pick feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer r0.5.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge.
#4337 opened Jun 12, 2026 by cuichenx Contributor Loading…
1
cp: docs(deepseek): update H100 Flash parallelism (4334) into r0.5.0 area:perf Performance optimizations and benchmarking cherry-pick docs Documentation-only updates or documentation debt docs-only With great power comes great responsibility. r0.5.0 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
#4336 opened Jun 12, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
cp: [examples] fix: Make Qwen2-Audio SFT runnable by default (4313) into r0.5.0 area:recipe Training recipes and launch configs bug Something isn't working cherry-pick r0.5.0 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
#4335 opened Jun 12, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
Add functional support matrix to README area:misc Cross-cutting utilities, logging, helpers, and other changes 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 r0.5.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge.
#4333 opened Jun 12, 2026 by snowmanwwg Contributor Loading…
5 tasks
3
[model] fix: adapt MCore dev attention gates area:model Model implementations and HF bridge logic bug Something isn't working full-test-suite needs-review PR is ready for code review and waiting on a reviewer
#4326 opened Jun 12, 2026 by yaoyu-33 Contributor Loading…
2
docs: Add initial fern migration area:misc Cross-cutting utilities, logging, helpers, and other changes docs Documentation-only updates or documentation debt needs-review PR is ready for code review and waiting on a reviewer
#4325 opened Jun 12, 2026 by chtruong814 Contributor Loading…
5 tasks
5
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-06-12) full-test-suite
#4324 opened Jun 12, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
fix(training): run K8s trainer from data-mover-synced code (r0.5.0 backport)
#4320 opened Jun 12, 2026 by ko3n1g Contributor • Draft
1
fix(training): run K8s trainer from data-mover-synced code area:training Training loop, callbacks, and runtime integration bug Something isn't working
#4319 opened Jun 12, 2026 by ko3n1g Contributor • Draft
1
26.06 perf summary 26.06 docs Documentation-only updates or documentation debt docs-only With great power comes great responsibility. r0.5.0 Auto-cherrypick to release branch. Apply before merge; cherrypick happens after merge.
#4318 opened Jun 12, 2026 by malay-nagda Contributor • Draft
5 tasks
1
Fix nemotronLabsDiffusion checkpoint conversion area:diffusion DFM module bug Something isn't working needs-more-tests Requires additional L0 and L1 test coverage before merge r0.5.0 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
#4317 opened Jun 12, 2026 by sajadn Contributor Loading…
5 tasks
10
refactor(data): move VLM sequence batching to collate
#4315 opened Jun 12, 2026 by yaoyu-33 Contributor • Draft
1
fix+feat(gemma2): fix SWA correctness bugs and add FlexAttention fused softcap+SWA path 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
#4308 opened Jun 11, 2026 by nvegesna-netizen Contributor Loading…
1
feat(data): add text chat collate for unified HF datasets area:data Dataset builders, preprocessing, and samplers 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
#4307 opened Jun 11, 2026 by yaoyu-33 Contributor Loading…
13
[model, feature] qwen3-omni: add packed sequence support and shared sequence utilities area:model Model implementations and HF bridge logic community-request feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer
#4304 opened Jun 11, 2026 by hbhflw2000 Contributor Loading…
1
chore(beep boop 🤖): Bump uv.lock (main, mcore-dev) (2026-06-11) full-test-suite
#4303 opened Jun 11, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
feat(model): thread weight_dtype through HF export for plain-dtype DeepSeek-V4 output area:model Model implementations and HF bridge logic feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer
#4301 opened Jun 11, 2026 by Meirtz Member Loading…
5
mirror: [model] fix: apply attention_value_scale to V in MiMo-V2-Flash 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
#4299 opened Jun 11, 2026 by ko3n1g Contributor Loading…
3
mirror: feat: Add EXAONE 4.0 model bridge (LG AI Research) area:model Model implementations and HF bridge logic community-request feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer
#4298 opened Jun 11, 2026 by ko3n1g Contributor Loading…
6
mirror: feat(model): add Gemma-4 E4B support (layer spec, checkpoint loader, parity check) area:model Model implementations and HF bridge logic feature New capabilities, enhancements, or enablement work ready-to-merge PR is approved, current, and only waiting for CI to pass before merge
#4297 opened Jun 11, 2026 by ko3n1g Contributor Loading…
7
mirror: fix(gemma4): map dense MLP pre-FFN norm to pre_feedforward_layernorm 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
#4296 opened Jun 11, 2026 by ko3n1g Contributor Loading…
3
refactor(docker): Dockerfile configurations and enhance environment variable management area:build Dependencies, packaging, images, and environment setup bug Something isn't working needs-review PR is ready for code review and waiting on a reviewer
#4288 opened Jun 10, 2026 by balasaajay Contributor Loading…
5 tasks
3
chore: hello-world stacked PR — layer 1
#4274 opened Jun 10, 2026 by ko3n1g Contributor • 1/3 • Draft
1
feat(diffusion): add LongLive WAN training path area:diffusion DFM module community-request feature New capabilities, enhancements, or enablement work needs-review PR is ready for code review and waiting on a reviewer
#4272 opened Jun 10, 2026 by AndysonYs Loading…
3 of 5 tasks
1
ProTip! Filter pull requests by the default branch with base:main.