Skip to content

Pull requests: NVIDIA-NeMo/Automodel

New pull request New
187 Open 2,478 Closed
187 Open 2,478 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(gpt-oss): route packed attention through THD
#3226 opened Jul 25, 2026 by akoumpa Contributor • Draft
2
fix(thd): derive packed padding mask from the pack layout, not token ids
#3223 opened Jul 25, 2026 by HuiyingLi Contributor Loading…
2 of 3 tasks
10
fix(glm): own the GLM MoE DSA config so qk_rope_head_dim survives
#3222 opened Jul 25, 2026 by HuiyingLi Contributor Loading…
2 of 3 tasks
6
perf(glm): adapt sparse MLA backward threads by block size community-request
#3220 opened Jul 25, 2026 by Butterfingrz Contributor Loading…
3 tasks done
1
feat(diffusion): integrate DMD2 into the native trainer
#3218 opened Jul 25, 2026 by jingyu-ml • Draft
3 tasks done
3
feat(diffusion): add Qwen-Image-Edit-2511 training support
#3217 opened Jul 24, 2026 by pthombre Contributor Loading…
3 tasks done
2
fix: preserve batch dimension in VLM validation community-request
#3216 opened Jul 24, 2026 by darth-pika-hu • Draft
3 tasks done
1
1
ci: reject new dynamic attribute access
#3212 opened Jul 24, 2026 by akoumpa Contributor • Draft
6 tasks done
2
feat: support pre-extracted frame sequence of video sample community-request
#3211 opened Jul 24, 2026 by GITsologun Loading…
2 of 3 tasks
1
8
fix(vlm): preserve mRoPE axes in PP chunking
#3208 opened Jul 24, 2026 by HuiyingLi Contributor Loading…
6
build: cache CUDA extension source builds
#3204 opened Jul 23, 2026 by akoumpa Contributor Loading…
18
fix(ci): omit token type IDs from vLLM parity prompts
#3198 opened Jul 23, 2026 by yuhezhang-ai Contributor Loading…
4 tasks done
4
chore: CI proxy for #3125
#3197 opened Jul 23, 2026 by HuiyingLi Contributor Loading…
3
chore: CI proxy for #3166
#3195 opened Jul 23, 2026 by HuiyingLi Contributor Loading…
3
chore: CI proxy for #3173
#3194 opened Jul 23, 2026 by HuiyingLi Contributor Loading…
4
feat(vlm): integrate packed CP and vision sharding for Qwen3.5-MoE
#3186 opened Jul 22, 2026 by yuhezhang-ai Contributor Loading…
6 of 9 tasks
5
feat: add ViSpec VLM draft training community-request waiting-on-maintainers Waiting on maintainers to respond
#3173 opened Jul 22, 2026 by khazic Contributor Loading…
1
8
ci: add Tulu-3 convergence + eval flow
#3171 opened Jul 22, 2026 by athitten Contributor Loading…
2 of 3 tasks
9
feat(speculative): add the multimodal speculative decoding (MSD) core community-request
#3166 opened Jul 22, 2026 by khazic Contributor Loading…
6 tasks done
9
feat(diffusion): add LTX-2.3 video+audio finetuning (training + prepr…
#3165 opened Jul 22, 2026 by linnanwang Contributor Loading…
3 tasks
6
perf(cuda-graphs): support selective AC and host-spilled MoE stash
#3164 opened Jul 22, 2026 by yaoyu-33 Contributor • Draft
1
feat(moe): drive router replay during activation-checkpointing recompute
#3156 opened Jul 21, 2026 by abhinavg4 • Draft
1
fix(bagel): initialize recipe config at boundary
#3155 opened Jul 21, 2026 by qiaochuz-nv Contributor Loading…
4 of 6 tasks
3
feat: add Kimi Linear model support
#3142 opened Jul 21, 2026 by HuiyingLi Contributor • Draft
37
fix(moe): preserve top-k routing under full AC
#3140 opened Jul 20, 2026 by yuhezhang-ai Contributor • Draft
3 tasks done
2
ProTip! Updated in the last three days: updated:>2026-07-22.