-
### 问题描述 Issue Description
因为官方不提供 ARM 架构对应 wheel 安装包,因此需要从源码进行手动编译。选取目标 PaddlePaddle 版本为最新 `develop` 分支。
1. cmake
执行命令:
> cmake .. -DPY_VERSION=3.8 -DPYTHON_EXECUTABLE=`which python3` -DWITH…
-
环境如下
系统:wsl2 + ubuntu 20.04
python版本:3.8
pytorch: 2.0.1
cuda: 11.7
出错运行代码如下,前面stage 0 - 4步运行都成功了
===================================
./run.sh --stage 5 --stop-stage 5
finetune...
rank: 0, pid…
-
There are already some old, half-baked Gatsby docs and we are writing some new ones.
- [ ] Make sure existing docs are up to date
- [x] Look at all the Gatsby docs
- [x] Structure - work out where ne…
-
多线程运行后发现内存泄漏,程序可以肯定的是已经delete掉LAC对象了,是我的用法有问题么?
at 0x4C2A1E3: operator new(unsigned long) (vg_replace_malloc.c:334)
==2286== by 0x124DE9BD: paddle::framework::details::OpInfoFiller::o
perator()(…
-
----------- Configuration Arguments -----------
_EOS_: 28270
batch_size: 256
channel1_num: 32
channel2_num: 16
data_path: ./data/data.pkl
emb_size: 200
ext_eval: False
learning_rate: 0.001
m…
-
相关环境:训练GPU v100, paddle1.4.1
重载环境:GPU v100, paddle1.5.0
背景:训练时只使用save_persistables保存为单个文件,现想重载权重文件进行评估和预测,但一直失败。
训练时使用
```
model_ckpt_name = 'params_pass_%05d.ckpt' % (epoch)
…
-
ONNX由torch模型export获得,面向任务为信息抽取。在转成paddle文件后使用方法如下:
```
with fluid.dygraph.guard():
for batch in text_idx:
text = fluid.dygraph.to_variable(np.array(batch))
_, outputs …
-
Open Video Success:C:/Users/Administrator/Desktop/1212.jpg
SubtitleArea:(375,401,240,709)
use GPU for acceleration
[Processing] start finding subtitles...
Exception in thread Thread-3:
Traceback …
-
有解决方案么?
报错信息如下:
Error: Can not import paddle core while this file exists: /home/***/.conda/envs/RocketQA/lib/python3.10/site-packages/paddle/fluid/libpaddle.so
Traceback (most recent call last):
…
-
我使用unsqueeze的示例程序能正确扩展维度:
x = fluid.layers.data(name='x', shape=[5, 10])
y = fluid.layers.unsqueeze(input=x, axes=[1])
但是当我使用自己读取的图片数据,在动态图中转变为variable之后,却总是报错:
with fluid.dygraph.guard…