Chang Lu

Could you add a colab notebook demo for your work please ?

2

Could you add a colab notebook demo for your work please ?

How to use it on bilibili?

2

How can i use it on bilibili?

我的数据集是有掩码和原图两部分组成的，训练的时候掩码需要和原图进行同样的变换，请问这个要怎么实现呢？

1 OPs are not supported yet : If

1

感谢您参与 X2Paddle 社区！ 问题模版为了 X2Paddle 能更好的迭代，例如新功能发布、 RoadMaps 和错误跟踪. :smile_cat: ## 问题描述 - 错误信息 INFO:root:paddle.__version__ = 2.2.2t. /Users/keter/opt/anaconda3/lib/python3.8/site-packages/onnx/mapping.py:27: DeprecationWarning: `np.object` is a deprecated alias for the builtin `object`. To silence this...

Bump paddlepaddle from 2.5.2 to 2.6.0

2

Bumps [paddlepaddle](https://github.com/paddlepaddle/paddle) from 2.5.2 to 2.6.0. Paddle 2.6 removed many apis in "fluid". There are bound to be a lot of bugs that the api won't find. We can change...

Add nhwc for maskrcnn

1

为 MaskRCNN 添加 nhwc 数据类型输入的支持 运行命令： ```bash # amp + nhwc FLAGS_cudnn_batchnorm_spatial_persistent=1 python tools/train.py -c configs/mask_rcnn/mask_rcnn_r50_1x_coco_nhwc.yml -o LearningRate.base_lr=0.0001 log_iter=1 use_gpu=True save_dir=./test_tipc/output/mask_rcnn_r50_1x_coco/benchmark_train/norm_train_gpus_0_autocast_fp32 epoch=1 TrainReader.batch_size=2 amp=True filename=mask_rcnn_r50_1x_coco TrainReader.shuffle=False --enable_ce=True # baseline python...

[Auto Parallel] Add Pipeline Memory Estimator

7

### PR Category Auto Parallel ### PR Types Improvements ### Description PipelineMemoryEstimator 类的主要功能是根据 program 来估计显存占用情况。在添加一个 program 的时候，会根据 program 来估计显存占用情况。与估计单个 program 的显存不同的是，PipelineMemoryEstimator 类还会维护一个 type_to_skip_gc_vars 字典，用来存储会被后续program使用的变量。 该类主要为 zb-vpp 自动调度服务，用于估计 program 的显存占用 测试环境，4卡1080Ti，Llama2，PaddleNLP...

[Auto Parallel] Add zero bubble vpp pipeline

5

### PR Category Auto Parallel ### PR Types New features ### Description 为 Paddle 添加 ZB VPP 编排方式 相关 Issue: - https://github.com/PaddlePaddle/Paddle/issues/62666 依赖 PR: - https://github.com/PaddlePaddle/Paddle/pull/63402

【Hackathon 6th No.44】add paddle rsqrt

6

Add `paddle.rsqrt` op for openvino paddle.rsqrt(x, name=None) The rsqrt activation function. The calculation formula is as follows: $$ out=\frac{1}{\sqrt{x}} $$  Waiting for: - https://github.com/openvinotoolkit/openvino/pull/23010