Paddle icon indicating copy to clipboard operation
Paddle copied to clipboard

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)

Results 1049 Paddle issues
Sort by recently updated
recently updated
newest added

### bug描述 Describe the Bug 基于Paddle develop分支 - PaddleDetection develop分支, 以及Paddle release2.3分支 - PaddleDetection release2.4分支,跑ppyolo网络,无论是4卡还是8卡,均会有50%的概率出现 loss=nan。环境和超参数修改如下: **原生参数8卡参数为(base_lr: 0.01,batch_size: 24)** - **出现loss=nan的8卡参数和脚本** 参数为(base_lr: 0.005,batch_size: 12),学习率和batch减半,训练脚本如下: `python -m paddle.distributed.launch --gpus 0,1,2,3,4,5,6,7 tools/train.py...

status/new-issue
type/bug-report

### PR types Bug fixes ### PR changes Others ### Describe Remove calibration file path when deploy quantize model

### 请提出你的问题 Please ask your question 我想同步每张卡上loss_buffer,每次只是同步某一行。但是同步之后,每张卡打印出来的变量不一样,是为什么呢? 代码和打印log如下: loss_buffer = paddle.zeros(shape=[100,4]) #[epoch,loss_num] paddle.distributed.barrier() paddle.distributed.all_reduce(loss_buffer[epoch_id]) print("loss_buffer[epoch_id]",loss_buffer[epoch_id]) ![image](https://user-images.githubusercontent.com/31760851/191503594-4e6f162a-5e4c-41e0-8cdd-7a5810879e09.png)

status/new-issue
type/question

### PR types Others ### PR changes Docs ### Describe 1. Delete some expression like 'This Op' 2. remove import numpy as np - cn docs pr: https://github.com/PaddlePaddle/docs/pull/5285

contributor
status: proposed

### PR types Bug fixes ### PR changes Others ### Describe [Dygraph] Fix bugs of mp in eager mode

### PR types Function optimization ### PR changes APIs ### Describe [Check]Enhance pooling output_size type check

### PR types Others ### PR changes Others ### Describe 当前yaml配置趋于完善,这个pr主要是删除前期由于机制不完善使用invoke配置的yaml及代码

### PR types Others ### PR changes Others ### Describe This PR support rsqrt_p in incubate

### PR types Others ### PR changes Others ### Describe sparse支持静态图: 1. 添加infer_meta:[#46016](https://github.com/PaddlePaddle/Paddle/pull/46016) 2. framework,pybind添加对SparseCooTensor的支持,后面再支持SparseCsrTensor 3. feed_op、fetch_op添加SparseCooTensor的支持,后面再支持SparseCsrTensor 4. 添加sparse_manual_op_sig.cc,sparse_manual_op.cc,当前主要添加3D点云模型中使用的算子: - sparse_coo_tensor,indices,values,to_dense, conv, relu, add 5. 当前先支持静态图的推理,后续再添加反向op。

### PR types Others ### PR changes Others ### Describe - Flake8 tracking issue: #46039 添加 remove-tabs pre-commit hook,以自动删除 tab,并且移除 Flake8 配置中 ignore 的 W191 和 E101 错误码 本 PR...

contributor