PaddlePaddle/Paddle

View on GitHub

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）

Stars: 24.1k
Forks: 6k
Open beginner issues: 0
Indexed issues: 993
Avg merge: 4d 17h
Merged PRs (30d): 60

Dominant language: C++
License: Apache-2.0
Last GitHub push: Sep 17, 2026
Latest indexed: Sep 19, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: good first issue

Beginner friendly Everything

Open All

993 open issues indexed

使用PPOCRv5训练模型，出现错误：Segmentation fault，再次训练会出现系统崩溃，无任何终端输出错误，直接CPU卡死。 Open

status/new-issue type/bug-report

Difficulty 5/5 Over a week Newbie friendliness 25/100

PaddlePaddle/Paddle#79758 · Opened Sep 10, 2026
Paddle Inference 3.3.1 OneDNN 执行时 ConvertPirAttribute2RuntimeAttribute 不支持 ArrayAttribute<DoubleAttribute> Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 74/100

PaddlePaddle/Paddle#79749 · Opened Sep 8, 2026
Conv2D 默认初始化(`Normal(0, √(2/fan_in))`)在无归一化深层/共享权重结构下激活指数放大，导致训练发散与 NaN 梯度 Open

Difficulty 4/5 3-5 days Newbie friendliness 45/100

PaddlePaddle/Paddle#79706 · 1 comment · Opened Aug 27, 2026
容器内 GPU MMU Fault: ENGINE GRAPHICS GPC1 GPCCLIENT_T1_3，导致推理回退到 CPU，重启容器后问题解决 Open

status/new-issue type/bug-report

Difficulty 4/5 3-5 days Newbie friendliness 35/100

PaddlePaddle/Paddle#79701 · Opened Aug 26, 2026
[Performance] cummax/cummin 的 CUDA kernel 硬编码 dim3(16,32)，长 innermost 维扫描比 torch.cummax 慢 19 倍、比 paddle.cumsum 慢 87 倍 Open

PaddlePaddle/Paddle#79670 · 1 comment · 1 assignee · Opened Aug 18, 2026
在windows上为rocm编译发生错误 Paddle only supports Python version >= 3.10 now Open

status/new-issue type/question

Difficulty 3/5 1-2 days Newbie friendliness 45/100

PaddlePaddle/Paddle#79637 · Opened Aug 8, 2026
paddle.nn.functional.silu backward returns -Inf for finite float64 gradient at x=-709 Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 52/100

PaddlePaddle/Paddle#79630 · Opened Aug 6, 2026
paddle.nn.functional.selu backward returns wrong finite gradient at x=-700 Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 55/100

PaddlePaddle/Paddle#79629 · Opened Aug 6, 2026
paddle.nn.functional.elu backward returns 0.0 for finite float64 gradient at x=-40 Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 56/100

PaddlePaddle/Paddle#79628 · Opened Aug 6, 2026
paddle.expm1 backward returns 0.0 for finite float64 gradient at x=-40 Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 52/100

PaddlePaddle/Paddle#79627 · Opened Aug 6, 2026
When will PaddlePaddle be available/kompatible for Python v3.14? Open

status/new-issue type/question

Difficulty 5/5 Over a week Newbie friendliness 25/100

PaddlePaddle/Paddle#79626 · 1 reaction · Opened Aug 6, 2026
paddle.slice_scatter segfaults instead of raising for length-mismatched control args Open

status/new-issue type/bug-report

Difficulty 4/5 3-5 days Newbie friendliness 55/100

PaddlePaddle/Paddle#79621 · Opened Aug 6, 2026
paddle.hypot overflows to Inf for large finite float64 inputs Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 52/100

PaddlePaddle/Paddle#79620 · Opened Aug 6, 2026
paddle.i1 prematurely overflows to Inf for finite float64 result Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 58/100

PaddlePaddle/Paddle#79619 · Opened Aug 6, 2026
paddle.i0 prematurely overflows to Inf for finite float64 result Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 55/100

PaddlePaddle/Paddle#79618 · 1 comment · Opened Aug 6, 2026
[Dy2St][CINN] Tensor kwargs 在第二次执行时导致 Embedding 读取错误 ID Open

status/new-issue type/bug-report

Difficulty 4/5 3-5 days Newbie friendliness 45/100

PaddlePaddle/Paddle#79580 · Opened Jul 29, 2026
[Dy2St] `full_graph=True` 下方法名与全局函数同名时调用被错误绑定 Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 72/100

PaddlePaddle/Paddle#79579 · Opened Jul 29, 2026
[PIR][CINN] `Tensor.view()` backward fails with `view_shape_grad` kernel output mismatch Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79559 · 1 assignee · Opened Jul 28, 2026
Support Python 3.14 for PaddlePaddle Open

status/new-issue type/feature-request

Difficulty 4/5 3-5 days Newbie friendliness 38/100

PaddlePaddle/Paddle#79527 · 1 comment · 8 reactions · Opened Jul 21, 2026
paddle.sparse.subtract() with CSR tensor containing negative column indices triggers heap-buffer-overflow Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79506 · 1 assignee · Opened Jul 18, 2026
paddle.nn.functional.embedding() with scalar out-of-range index triggers heap-buffer-overflow Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79505 · 1 assignee · Opened Jul 18, 2026
paddle.nn.functional.conv3d_transpose() triggers heap-buffer-overflow for NDHWC input with grouped convolution parameters Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79504 · 1 assignee · Opened Jul 18, 2026
paddle.linalg.eigvals() crashes on float32 matrix containing inf values instead of raising a clean exception Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79503 · 1 assignee · Opened Jul 18, 2026
paddle.eye() with negative 0-D Tensor dimensions triggers heap-buffer-overflow in EyeFunctor Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79502 · 1 assignee · Opened Jul 18, 2026
paddle.tensordot() with scalar float16 inputs triggers stack-buffer-overflow in DDim::operator[] Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79501 · 1 assignee · Opened Jul 18, 2026
[Autograd] Missing higher-order gradients block molecular force training Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

PaddlePaddle/Paddle#79493 · Opened Jul 16, 2026
Heap buffer overflow in `DenseTensor` stream deserialization (LoD length integer-division truncation) Open

status/new-issue type/bug-report

Difficulty 4/5 3-5 days Newbie friendliness 55/100

PaddlePaddle/Paddle#79492 · 2 comments · Opened Jul 16, 2026
Out-of-bounds read in the CPU embedding kernel when `padding_idx` is set (heap memory disclosure + remote crash) Open

status/new-issue type/bug-report

Difficulty 3/5 1-2 days Newbie friendliness 68/100

PaddlePaddle/Paddle#79491 · 3 comments · Opened Jul 16, 2026
Remote Code Execution via Unsafe Pickle Deserialization in the CIFAR dataset loader (`paddle.vision.datasets.Cifar10` / `Cifar100`) Open

status/new-issue type/bug-report

Difficulty 4/5 3-5 days Newbie friendliness 52/100

PaddlePaddle/Paddle#79490 · 3 comments · Opened Jul 16, 2026
`paddle.static.nn.while_loop` with to_static(full_graph=True) segfaults when body captures an eager tensor Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79477 · 1 assignee · Opened Jul 14, 2026
`paddle.static.nn.case\paddle.static.nn.switch_case` with to_static(full_graph=True) segfaults when branch functions capture eager tensors Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79476 · 1 assignee · Opened Jul 14, 2026
`paddle.unique` on an empty GPU tensor aborts under to_static when return_inverse=True Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79475 · 1 assignee · Opened Jul 14, 2026
`paddle.mode ` on an empty GPU tensor aborts the Python process Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79474 · 1 assignee · Opened Jul 14, 2026
is these a implement issue for saslprep Open

status/new-issue type/question

PaddlePaddle/Paddle#79472 · 1 assignee · Opened Jul 14, 2026
Paddle autograd returns NaN gradients for inactive paddle.where branch with sqrt Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79464 · 1 assignee · Opened Jul 12, 2026
paddle.exp returns NaN imaginary part for complex Inf input Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79463 · 1 assignee · Opened Jul 12, 2026
paddle.logaddexp incorrectly returns NaN for same-sign infinity inputs Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79462 · 1 assignee · Opened Jul 12, 2026
paddle.nn.functional.gelu overflows to Inf for maximum float32 input Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79461 · 1 assignee · Opened Jul 12, 2026
paddle.erfinv incorrectly returns Inf for inputs outside the valid domain Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79460 · 1 assignee · Opened Jul 12, 2026
paddle.heaviside incorrectly returns 0.0 for NaN input Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79459 · 1 assignee · Opened Jul 12, 2026
paddle.digamma incorrectly returns Inf instead of NaN for zero and tiny float32 inputs Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79458 · 1 assignee · Opened Jul 12, 2026
项目无法用A800启动 Open

status/new-issue type/others

PaddlePaddle/Paddle#79452 · 1 comment · 1 assignee · Opened Jul 10, 2026
PIR-TensorRT converter cannot convert any PP-OCRv5/PP-StructureV3 server model (multiple failure classes; after local crash fixes, engines build cleanly but emit EMPTY output; stable 3.3.1 + develop, TRT 10.13/10.16, sbsa/SM121) Open

PaddlePaddle/Paddle#79449 · 1 comment · 1 assignee · Opened Jul 9, 2026
[Call for Bench] SWE-Paddle 社区共建 Open

PaddlePaddle/Paddle#79447 · 4 reactions · 1 assignee · Opened Jul 9, 2026
Enable CI Support for ppc64le (IBM Power) Architecture Open

PaddlePaddle/Paddle#79436 · 1 comment · 1 assignee · Opened Jul 8, 2026
pocketfft gitlab仓库疑似删库跑路 Cmake一配置就爆炸 Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79415 · 1 comment · 1 assignee · Opened Jul 3, 2026
Windows pip wheels for paddlepaddle-gpu 3.2.2/3.3.1 warn that Paddle was compiled with cuDNN 9.9 while installed dependency provides cuDNN 9.5.1.17 Open

PaddlePaddle/Paddle#79388 · 1 assignee · Opened Jun 28, 2026
[GPU] paddle.sort and paddle.argsort crash on zero-batch tensors Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79367 · 1 assignee · Opened Jun 24, 2026
[CPU/GPU] paddle.tile rejects repeat_times containing 0 Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79366 · 1 assignee · Opened Jun 24, 2026
[CPU/GPU] Elementwise binary ops fail on broadcastable zero-sized tensors Open

status/new-issue type/bug-report

PaddlePaddle/Paddle#79365 · 1 assignee · Opened Jun 24, 2026

PaddlePaddle/Paddle

Get new issues in your inbox