Skip to content

Pull requests: modelscope/ms-swift

New pull request New
64 Open 3,235 Closed
64 Open 3,235 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] change default values of topk/topp
#8442 opened Mar 26, 2026 by hjh0119 Loading…
1
Add loss scale from data
#8430 opened Mar 25, 2026 by hpsun1109 Loading…
5
[bugfix] fix megatron opsd with top-k
#8428 opened Mar 25, 2026 by hjh0119 Loading…
3
REAL Loss (Rewards as Labels) for GRPO Training
#8424 opened Mar 25, 2026 by li2zhi Loading…
2 of 4 tasks
@hjh0119
9
[megatron] Refactor mcore bridge
#8422 opened Mar 25, 2026 by Jintao-Huang Loading…
4
[bugfix] initialize step3 image inputs for text-only requests
#8421 opened Mar 24, 2026 by mshicom Loading…
3
[bugfix] harden llm infer streaming and runtime parsing
#8418 opened Mar 24, 2026 by mshicom Loading…
1
[bugfix] fallback to eager attention for MPS VL loaders
#8419 opened Mar 24, 2026 by mshicom Loading…
3
Fix/environment variable in multinode train
#8413 opened Mar 24, 2026 by huangfu170 Loading…
3 tasks
3
Fix wandb logging step and metrics reporting
#8402 opened Mar 23, 2026 by SundayVHan Loading…
3 tasks
2
[megatron]support hybridep, Add moe_flex_dispatcher_backend argument and hybridep padding fix
#8393 opened Mar 20, 2026 by tingmingzhong Loading…
4
[megatron] support multimodal MTP
#8390 opened Mar 20, 2026 by Jintao-Huang Loading…
1
[megatron]add megatron log
#8348 opened Mar 16, 2026 by yangbofun Loading…
1 of 4 tasks
3
[megatron] support the fake distributed process group
#8347 opened Mar 16, 2026 by yangbofun Loading…
1 of 4 tasks
5
[bugfix] Fix abnormal grad_norm under GRPO LoRA + DeepSpeed ZeRO-0 (fix #6815)
#8341 opened Mar 14, 2026 by alphadl Loading…
4
[bugfix] Web UI: load from checkpoint path by type instead of dirty exception check
#8340 opened Mar 14, 2026 by alphadl Loading…
3
[bugfix] fix load audio from local video path when USE_AUDIO_IN_VIDEO (fix #8332)
#8339 opened Mar 14, 2026 by alphadl Loading…
1
4
[bugfix] fix PeftModel save_pretrained selected_adapters when using custom adapter_name
#8337 opened Mar 14, 2026 by alphadl Loading…
4
feat(template): add MiniCPM-o-4.5 training template with audio support and fix image_bound bug
#8307 opened Mar 12, 2026 by xxddccaa Loading…
5
SymPO
#8245 opened Mar 9, 2026 by JiangWu0826 Loading…
1 of 4 tasks
1
[model] support kimi-k2.5 transformers & megatron
#8239 opened Mar 8, 2026 by Jintao-Huang Loading…
3
Feature/ms swift custom
#8222 opened Mar 6, 2026 by LEWISZZZcc Loading…
4 tasks
18
[WIP] Moe kernel for qwen3 omni in ascend
#8214 opened Mar 5, 2026 by jiaqiw09 Loading…
1 of 4 tasks
3
[megatron]feat: Add routing replay support for Megatron-Swift GRPO
#8196 opened Mar 4, 2026 by XianlongLi Loading…
1 of 4 tasks
@hjh0119
24
feat: log grpo input images to wandb
#8157 opened Mar 2, 2026 by shunk031 Loading…
1 of 4 tasks
4
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.