-
Notifications
You must be signed in to change notification settings - Fork 178
Pull requests: ByteDance-Seed/VeOmni
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix: NPU Hang/Deadlock during DTensor parameter loading in FSDP2
ascend
everything about Ascend support
fix
#642
opened Apr 10, 2026 by
First-Frost-code
Loading…
[WIP]: support npu qwen3p5 and qwen3p5-vl
ascend
everything about Ascend support
wip
#641
opened Apr 10, 2026 by
yicheng-gong
Contributor
Loading…
[arguments]fix bug for args.train.accelerator.fsdp_config.mixed_precision.enable
bug
Something isn't working
fix
#638
opened Apr 9, 2026 by
UserChen666
Loading…
[misc] fix: revert fsdp basic modules, fix it with VL models only
fix
misc
Every misc
#634
opened Apr 8, 2026 by
heidongxianhua
Contributor
Loading…
[docker] feat: update to torch2.11 + cu130
docker
#629
opened Apr 2, 2026 by
FoolPlayer
Collaborator
Loading…
[model]feat: add NPU support for Qwen3.5
ascend
everything about Ascend support
#628
opened Apr 2, 2026 by
yanghw116
Loading…
[model]feat: Qwen3.5 is compatible with NPU
ascend
everything about Ascend support
#600
opened Mar 23, 2026 by
wang-hua-2019
Contributor
Loading…
[model] feat: [transformers-v5] Introduce new registration based kernel replacement.
#569
opened Mar 16, 2026 by
piyifan123
Collaborator
Loading…
[model] feat: [transformers v5] support qwen3vl for transformer v5
hf_v5
Related for transformers v5
#527
opened Mar 2, 2026 by
yiwzhao
Collaborator
Loading…
[parallel] feat: Vision Data Parallel — O(1) communication alternative to patch-level SP
#505
opened Feb 24, 2026 by
aoshen524
Loading…
2 of 3 tasks
[models] chore: Change transformers v5 support for qwen3_moe to use HF v5 style expert weight layout and add a converter impl.
hf_v5
Related for transformers v5
misc
Every misc
#500
opened Feb 24, 2026 by
piyifan123
Collaborator
•
Draft
[task] feat: support sequence classification tasks
#470
opened Feb 11, 2026 by
yiwzhao
Collaborator
Loading…
[model] fix: Incorrect usage of the 'check_model_inputs' decorator
fix
#457
opened Feb 5, 2026 by
HSYZhang
Contributor
Loading…
6 tasks done
[misc] chore: add_copy_right
misc
Every misc
#438
opened Jan 30, 2026 by
FoolPlayer
Collaborator
Loading…
Draft [models] feat: Add a modeling patch gen sample for qwen3
#424
opened Jan 26, 2026 by
piyifan123
Collaborator
Loading…
6 tasks
[docs]Update ascend_quick_start doc
ascend
everything about Ascend support
#225
opened Nov 27, 2025 by
Alter-A1ways
Loading…
4 of 6 tasks
[optim, config] feat: add support for Muon optimizer via dion
doc
Improvements or additions to documentation
#216
opened Nov 25, 2025 by
clarkipeng
Loading…
[fix] [model] auto-patch all Attention layers to ensure cu_seq_lens stays on CPU for NPU fused-attention.
ascend
everything about Ascend support
#199
opened Nov 17, 2025 by
A1waysBeenHere
Contributor
Loading…
4 of 6 tasks
Add TensorBoard support for training metrics logging
#195
opened Nov 14, 2025 by
iqiancheng
Contributor
Loading…
train qwen3-vl-moe on ShareGPT4V-small with quick-start
#194
opened Nov 14, 2025 by
iqiancheng
Contributor
Loading…
Previous Next
ProTip!
Type g i on any issue or pull request to go back to the issue listing page.