-
Notifications
You must be signed in to change notification settings - Fork 184
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
【Hackathon 5th No.61】Skillful nowcasting of extreme precipitation with NowcastNet #593
Conversation
Thanks for your contribution! |
input_frames = frames[:, : self.input_length] | ||
input_frames = input_frames.reshape((batch, self.input_length, height, width)) | ||
# Evolution Network | ||
intensity, motion = self.evo_net(input_frames) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
发现推理结果的图片与参考代码的有些差异,应该是某处没有对齐导致的,我这边帮忙查了下发现这个网络给定相同输入时输出存在2e-4左右的差异,需要在详细查一下,修改后建议再对整个网络整体检查下对齐情况
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
从其他模型复现遇到的问题比较,可能是初始化随机数不同结果有差异,如果是网络问题会模型不能运行或结果差异会较大。
模型是按源码复现的,那如果不大明确,就不在这上面花时间了
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
当前情况:paddle的grid_sample的CPU结果跟torch的GPU结果能对齐,GPU结果跟torch的CPU结果能对齐,但是把这个运算改到cpu上,后面仍然有没对齐的地方。by @HydrogenSulfate @zhiminzhang0830
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
后面仍然有没对齐
@luotao1 @HydrogenSulfate 这边check过了,后面没对齐仍然是grid_sample的问题,这个PR除了grid_sample外基本OK了,还有几处格式问题需要辛苦 @co63oc 再优化一下就可以合入了。
|
||
|
||
class Generative_Decoder(paddle.nn.Layer): | ||
def __init__(self, opt): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Generative_Decoder 可以直接改为使用多个入参么,类似这样
def __init__(self, ngf, evo_ic, gen_oc, ic_feature):
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
其他子层需要使用opt,如果拆开其他子层也要拆开,代码参数变量较多
NowCastNet需要用到 OpenCV, 需要在requirements.txt里添加该库 opencv-python |
已修改 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
…h NowcastNet (PaddlePaddle#593) * Add nowcastnet * Fix
PR types
Others
PR changes
Others
Describe
No.61:Skillful nowcasting of extreme precipitation with NowcastNet
论文链接:
https://www.nature.com/articles/s41586-023-06184-4#Abs1
参考代码链接:
https://codeocean.com/capsule/3935105/tree/v1
RFC PaddlePaddle/community#705
使用预训练模型生成图
torch测试图
paddle测试图
运行命令
预训练模型和数据集下载