Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
117 Open 1,329 Closed
117 Open 1,329 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: prioritize dataclass recursion in config conversion community-request
#1878 opened Jan 8, 2026 by yufoo1 Loading…
3 of 5 tasks
1
Only put ddp init in a synchronized cuda side-stream
#1877 opened Jan 8, 2026 by jstjohn Loading…
5 tasks
2
chore: Add evaluation pipeline
#1876 opened Jan 7, 2026 by ko3n1g Loading…
5 tasks
chore(beep boop 🤖): Bump uv.lock (r0.2.0) (2026-01-07)
#1873 opened Jan 7, 2026 by ko3n1g Loading…
1 task
chore(beep boop 🤖): Bump uv.lock (main) (2026-01-07)
#1872 opened Jan 7, 2026 by ko3n1g Loading…
1 task
Update Deepseek V3 B200 recipe community-request
#1869 opened Jan 6, 2026 by dingqingy-nv Loading…
2
Fix Qwen3-VL packed RoPE offsets and MoE export merging
#1868 opened Jan 6, 2026 by yaoyu-33 Loading…
9
Enable THD context-parallel packed sequences
#1867 opened Jan 6, 2026 by yaoyu-33 Loading…
14
fix dict not dataclass issue for nvrx-straggler
#1862 opened Jan 6, 2026 by JeffPengCoder Loading…
3 tasks
2
[WIP] padded_vocab_size support community-request
#1854 opened Jan 5, 2026 by lostkevin • Draft
3 of 5 tasks
1
[PEFT] feat: Add support for MoE router weight adapter wrapper
#1834 opened Dec 31, 2025 by HollowMan6 Loading…
2 of 5 tasks
8
Malay/tmp use cudnn ln
#1825 opened Dec 30, 2025 by malay-nagda • Draft
5 tasks
1
fix: correct throughput calculation with sequence packing and warn on missing checkpoint community-request
#1822 opened Dec 29, 2025 by yurekami Loading…
3 tasks
1
11
Revise README with quickstart instructions
#1820 opened Dec 29, 2025 by snowmanwwg Loading…
5 tasks
Sarvam MoE support community-request
#1814 opened Dec 28, 2025 by mohit-sarvam • Draft
5 tasks
3
feat: add distributed save to speed up save hf weights community-request
#1806 opened Dec 24, 2025 by LiuXTao Loading…
5 tasks
6
Add Qwen3-VL support with Megatron-FSDP
#1801 opened Dec 24, 2025 by xuwchen Loading…
5 tasks
4
Add GLM4.5 V model
#1798 opened Dec 23, 2025 by suiyoubi Loading…
5 tasks
5
disable moe a2a
#1797 opened Dec 23, 2025 by malay-nagda • Draft
5 tasks
1
PyTorch profiler interface for performance recipes to reduce profiling overhead community-request
#1789 opened Dec 22, 2025 by dingqingy-nv Loading…
14
fix: Detach the params to avoid slow down training community-request
#1777 opened Dec 19, 2025 by moehanabi Loading…
1 of 5 tasks
3
QWen3-VL Performance Test
#1769 opened Dec 18, 2025 by shifangx Loading…
5 tasks
9
Peft Bridge
#1766 opened Dec 18, 2025 by yaoyu-33 Loading…
5 tasks
23
Improving Qwen3-235B-A22B B200 performance. community-request performance
#1760 opened Dec 17, 2025 by dingqingy-nv Loading…
4
feat: Validate PEFT target modules
#1747 opened Dec 16, 2025 by yaoyu-33 Loading…
2
ProTip! Filter pull requests by the default branch with base:main.