Skip to content

Pull requests: NVIDIA-NeMo/Megatron-Bridge

New pull request New
138 Open 1,444 Closed
138 Open 1,444 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

[misc] refactor: Replace deprecated typing imports (#993) community-request
#2057 opened Jan 24, 2026 by KunalSachdev2005 Loading…
1 of 5 tasks
1
1
ci: Apply save_config_filepath override
#2054 opened Jan 24, 2026 by ko3n1g • Draft
5 tasks
1
Refactor provider_bridge for Llama and Qwen models
#2052 opened Jan 23, 2026 by yaoyu-33 • Draft
10 of 16 tasks
1
Add Gemma3 VL example
#2049 opened Jan 23, 2026 by meatybobby Loading…
5 tasks
10
fix hf tokenizers path
#2047 opened Jan 23, 2026 by dimapihtar Loading…
5 tasks
4
[doc] feat: Document adapter merge verification in stream_adapter_weights example
#2042 opened Jan 23, 2026 by yaoyu-33 Loading…
1
refactor(mimo): unified MimoModelProvider with ModuleSpec-based API
#2040 opened Jan 22, 2026 by aroshanghias-nvd Loading…
11
fix: Handle init=False dataclass fields for config loading
#2034 opened Jan 22, 2026 by ananthsub Loading…
2 of 5 tasks
3
[model] feat: Model Support for Ling MoE V2 Model community-request
#2028 opened Jan 22, 2026 by ccclyu Loading…
3 of 5 tasks
2
disable precision_aware_optimizer for FSDP bug Something isn't working performance/release Performance items related with NeMo release performance
#2025 opened Jan 22, 2026 by malay-nagda • Draft
5 tasks
26.02
@malay-nagda
1
add peft to recipe qwen3vl
#2023 opened Jan 22, 2026 by yashaswikarnati Loading…
5 tasks
1
fix: fix exported model.layers.*.self_attn.rotary_emb.inv_freq shape community-request
#2021 opened Jan 22, 2026 by conver334 Loading…
2 of 5 tasks
3
Fix vp check and add checkpoint configuration support performance/release Performance items related with NeMo release performance
#2020 opened Jan 22, 2026 by rsalagame-nvidia Loading… 26.02
@rsalagame-nvidia
14
Proposal for more scalable seq-pack data format
#2019 opened Jan 21, 2026 by marcromeyn • Draft
5 tasks
1
readme_perf_script_example
#2017 opened Jan 21, 2026 by malay-nagda Loading…
5 tasks
4
[M4]: Add M4 end2end support and qwen3 examples
#2011 opened Jan 21, 2026 by yaoyu-33 Loading…
16
Enable thd attention in gpt-oss recipe + add recipe tests
#2009 opened Jan 21, 2026 by cuichenx Loading…
5 tasks
4
feat(mimo): phase 3 - data loading
#2007 opened Jan 20, 2026 by aroshanghias-nvd Loading…
1 of 5 tasks
1
feat(mimo): phase 2 - model provider, DDP wrapping, process groups
#2004 opened Jan 20, 2026 by aroshanghias-nvd Loading…
2 of 4 tasks
1
[Draft] THD training in VLMs
#1997 opened Jan 20, 2026 by cuichenx Loading…
11 tasks
Support direct export of block-wise FP8 weights and scaling factors community-request
#1994 opened Jan 20, 2026 by eternally-z Loading…
5 tasks
5
remove legacy tokenizer system
#1989 opened Jan 19, 2026 by dimapihtar • Draft
5 tasks
1
feat: Add video support for Qwen3-VL model community-request
#1968 opened Jan 16, 2026 by pedrohenriqueamartins Loading…
2
Make TransformerConfig an attribute in MambaProvider
#1962 opened Jan 15, 2026 by maanug-nv • Draft
5 tasks
1
Remove deprecated mamba2 model configs
#1961 opened Jan 15, 2026 by ananthsub Loading…
2 of 5 tasks
7
ProTip! no:milestone will show everything without a milestone.