paddlepaddle / contrib Goto Github PK

您好，我想问下，我运行danet的train_executor.py，用GPU运行怎么修改代码？可以帮忙解答下吗，谢谢了。

1. 运行pip install -r requirement.txt有以下报错
   ERROR: Could not find a version that satisfies the requirement cuda>=8.0 (from -r requirements.txt (line 5)) (from versions: none)
   ERROR: No matching distribution found for cuda>=8.0 (from -r requirements.txt (line 5))

您好，请问一下，我如果用pytorch复现您的danet,需要修改那些地方？希望有空帮忙解答一下，谢谢了。

Hi, the pre-trained model "DANet101_better_model_paddle1.6" can not be loaded successfully in dygraph mode.

import paddle.fluid as fluid
path = 'DANet_models/DANet101_better_model_paddle1.6'
with fluid.dygraph.guard():
    state_dict = fluid.load_dygraph(path)
    print(state_dict)

meet the error

_pickle.UnpicklingError: invalid load key, '\x02'.

Environments:
python3.6.10
paddlepaddle-gpu 1.8.3.post107

您好，非常感谢您的项目，但是我有个疑问。您写了动态图版本的程序，一般我们会再训练时使用resnet当预训练模型，但是我在网上没有找到resnet的动态图预训练模型。请问您在训练的时候，是怎么处理的？还是说您这边有预训练模型，方便给个下载链接吗？

哈哈哈