Skip to content

Pull requests: PaddlePaddle/Paddle

New pull request New
504 Open 58,605 Closed
504 Open 58,605 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

[XPU] Fix fused_rotary_position_embedding precision: skip zero-numel k/v tensor
#78636 opened Apr 10, 2026 by YqGe585 Member Loading…
1
[XPU] Fix bfloat16 FC ops precision: use fp32 accumulation to match GPU
#78635 opened Apr 10, 2026 by YqGe585 Member Loading…
2
20260407 add ai edited test
#78634 opened Apr 10, 2026 by liuhao2638 Contributor Loading…
1
[Cpp API Compatibility] Align indexless CUDA compat ops with current device
#78631 opened Apr 10, 2026 by SigureMo Member Loading…
18
[XPU] add fused_dropout_add XPU kernel and remove Python fallback
#78629 opened Apr 10, 2026 by YqGe585 Member Loading…
1
[Cherry-pick][Operator] Align CPU randperm_kernel with PyTorch for bit-exact output #78614
#78627 opened Apr 10, 2026 by zhengshengning Contributor Loading…
2
[XPU] fix pinned memory audo transform XPU
#78626 opened Apr 10, 2026 by lj970926 Contributor Loading…
1
[XPU] fix float32 matmul precision: use FC_FLOAT default instead of FC_TF32 on XRE5
#78625 opened Apr 10, 2026 by YqGe585 Member Loading…
2
[LargeTensor][phi_ops] Fix int32 overflow in memory_efficient_attention, conv_transpose, and class_center_sample kernels
#78624 opened Apr 9, 2026 by DanielSun11 Contributor Loading…
2
[Precision Depth Alignment]fix p_norm_grad patch
#78623 opened Apr 9, 2026 by AlAuAu Contributor Loading…
2
[cherry-pick] Fix put_along_axis grad kernel crash for integer div-by-zero…
#78621 opened Apr 9, 2026 by wangbingguang2026 Contributor Loading…
3
add flashattn build for arm
#78619 opened Apr 9, 2026 by liuhao2638 Contributor Loading…
1
[DO NOT MERGE] test fleet ai unit
#78618 opened Apr 9, 2026 by tianlef Contributor Loading…
1
[API compatibility] update paddle.Tensor.exp_ and paddle.exp_ API Compatibility contributor External developers
#78617 opened Apr 9, 2026 by Le-soleile Contributor Loading…
6
[Operator] Align CPU randperm_kernel with PyTorch for bit-exact output
#78614 opened Apr 8, 2026 by zhengshengning Contributor Loading…
5
[AP] Add support for custom device.
#78610 opened Apr 8, 2026 by Xreki Contributor Loading…
2
[Performance Optimization] Add fast md5sum
#78602 opened Apr 8, 2026 by ForFishes Member Loading…
3
[API Compatibility] sgn/signbit/tensordot/tril_indices/vander/logaddexp/moveaxis/nan_to_num/nanmean/nansum/*scatter Edit By AI Agent API Compatibility
#78600 opened Apr 7, 2026 by zhwesky2010 Contributor Loading…
2
feat(HIP): register bfloat16 kernels for conv2d/conv3d/depthwise_conv2d on ROCm contributor External developers PaddlePaddle Hackathon
#78587 opened Apr 4, 2026 by fchange Loading…
@luotao1 @yongqiangma
4
[release/3.3][RMSNorm] remove invvar from rms_norm Python API return value contributor External developers
#78578 opened Apr 3, 2026 by ShigureNyako Contributor Loading…
2
【FSDP】Add FSDP in dynamic
#78577 opened Apr 3, 2026 by Xing-lil Contributor Loading…
13
[API Compatibility] Support legacy reduction args for BCE/MSE/multi_margin/nll_loss API Compatibility contributor External developers
#78574 opened Apr 3, 2026 by Aidenwu0209 Contributor Loading…
9
[Paddle Pilot] Align cross_entropy precision for soft-label, label-smoothing, and ND reduction under FLAGS_use_accuracy_compatible_kernel
#78572 opened Apr 3, 2026 by zrr1999 Member Loading…
2
[API Compatibility] Support arg closure for paddle.optimizer.optimizer.step API Compatibility contributor External developers
#78570 opened Apr 2, 2026 by algorithm1832 Contributor Loading…
11
fix cp rank_id bug
#78564 opened Apr 2, 2026 by Eddie-Wang1120 Contributor Loading…
3
ProTip! Add no:assignee to see everything that’s not assigned.