Search results

[ROCm] HIP BF16 仍缺失 layer_norm / softmax 算子与 conv2d_add 融合 pass，PaddleOCR-VL 等模型无法在 AMD GPU 上完整 BF16 推理

Status: Open.

#78710 In PaddlePaddle/Paddle;

· austin1997 opened on Apr 18, 2026

Is the paddlepaddle package on pypi official?

Status: Open.

#78706 In PaddlePaddle/Paddle;

· Frojdholm opened on Apr 17, 2026

[ROCm] HIP BF16 Precision Type Not Supported on AMD GPU

Status: Open.

#78664 In PaddlePaddle/Paddle;

· Lau-JW opened on Apr 13, 2026

关于安装paddlepaddle3.3.1版本训练pp_liteseg模型报错

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78663 In PaddlePaddle/Paddle;

· Hold-on-li opened on Apr 13, 2026

[ARM64][Segmentation Fault] PaddleOCR crashes on Kirin 9006C / UOS 1070 with kernel 5.4.96

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78616 In PaddlePaddle/Paddle;

· panyvtao opened on Apr 9, 2026

[Bug] macOS 26 (Sequoia 15) support broken in paddlepaddle 3.3.1

Status: Open.

#78542 In PaddlePaddle/Paddle;

· edisoncccc opened on Apr 1, 2026

【HACKATHON 10th ERNIE with Partners】文心合作伙伴赛道

PaddlePaddle Hackathon

Status: Open.

#78485 In PaddlePaddle/Paddle;

· onecatcn opened on Mar 25, 2026

paddle.logsumexp 对输入维度限制为 <=4

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78469 In PaddlePaddle/Paddle;

· tingPetty opened on Mar 25, 2026

paddle.greater_equal 在 int32 与 float32 比较时类型提升报错

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78468 In PaddlePaddle/Paddle;

· tingPetty opened on Mar 25, 2026

[CPU] 多个 Paddle 算子在 float16 输入下报 NotFound kernel

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78466 In PaddlePaddle/Paddle;

· tingPetty opened on Mar 25, 2026

paddle.shape 不支持10维 Tensor，抛出 Unimplemented error

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78463 In PaddlePaddle/Paddle;

· tingPetty opened on Mar 24, 2026

paddle.masked_fill 在整型输入场景出现 int32/int64 类型读取不一致报错

status/new-issue新建新建

type/bug-report报bug报bug

Status: Open.

#78462 In PaddlePaddle/Paddle;

· tingPetty opened on Mar 24, 2026