AWAN icon indicating copy to clipboard operation
AWAN copied to clipboard

Train data

Open jovian-2wang opened this issue 4 years ago • 18 comments

Hello, can you share your training data set? I can't find this part of the data set from the Internet. Thank you

jovian-2wang avatar Jul 03 '20 02:07 jovian-2wang

You can get it in the participate tab of the link "https://competitions.codalab.org/competitions/22225".

Deep-imagelab avatar Jul 03 '20 02:07 Deep-imagelab

Hello,running clean / test / valid_ result.py ,whether you are running out of memory? Thanks.

msn321 avatar Sep 18 '20 11:09 msn321

@jovian-2wang Hello,what are the results of running train_data_preprocess.py and valid_data_preprocess.py? I feel my fault.Thank you.

msn321 avatar Sep 24 '20 12:09 msn321

@Deep-imagelab Hello, when I was running main.py file, I encountered the following problem. I searched for information and tried several solutions, but failed to run it successfully. Do you know the reason?How can it be modified? Thank you

Problem: D:\software\anaconda3\envs\pytorch\python.exe E:/IPIC-AWAN-master/AWAN_Clean/train/main.py

loading dataset ... Train1:0,Train2:0,Train3:0,Train4:0, Validation set samples: 0

building models_baseline ... Parameters number is 17461617 Traceback (most recent call last): File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 186, in main() File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 95, in main train_loss, iteration, lr = train(train_loader, model, criterion_train, optimizer, epoch, iteration, opt.init_lr, opt.decay_power, opt.trade_off) File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 145, in train return losses.avg, iteration, lr UnboundLocalError: local variable 'lr' referenced before assignment

msn321 avatar Sep 29 '20 07:09 msn321

@Deep-imagelab Hello, when I was running main.py file, I encountered the following problem. I searched for information and tried several solutions, but failed to run it successfully. Do you know the reason?How can it be modified? Thank you

Problem: D:\software\anaconda3\envs\pytorch\python.exe E:/IPIC-AWAN-master/AWAN_Clean/train/main.py

loading dataset ... Train1:0,Train2:0,Train3:0,Train4:0, Validation set samples: 0

building models_baseline ... Parameters number is 17461617 Traceback (most recent call last): File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 186, in main() File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 95, in main train_loss, iteration, lr = train(train_loader, model, criterion_train, optimizer, epoch, iteration, opt.init_lr, opt.decay_power, opt.trade_off) File "E:/IPIC-AWAN-master/AWAN_Clean/train/main.py", line 145, in train return losses.avg, iteration, lr UnboundLocalError: local variable 'lr' referenced before assignment

Please run data pre-processing firstly.

Deep-imagelab avatar Sep 29 '20 08:09 Deep-imagelab

I know, I follow your README, but can't

msn321 avatar Sep 29 '20 09:09 msn321

Hello, while the paper shows MRAE 0.06+ on Real world NTIRE 2020, I use AWAN+ algorithm to test the validation set (provided by competition in google drive)on real word but get MRAE 0.1+. Do you have any different tricks that not published in your paper?

Fern-Luiz avatar Apr 15 '21 08:04 Fern-Luiz

您可以在“https://competitions.codalab.org/competitions/22225”链接的参与选项卡中获取它。

这个比赛已经结束了,这个链接也无法再下载数据了,请问有保存当时训练和验证的数据吗?如果有的话,可否通过网盘链接分享一下?非常感谢您杰出的工作!

JackWhite-rwx avatar Sep 30 '22 03:09 JackWhite-rwx

我把数据传好了,有效期是30天,你们可以尽快下载下。

链接:https://pan.baidu.com/s/1BjvPWlXbjGxmuUZmH5tnGw

提取码:2022

在 2022-09-30 15:02:45,"img_wcx1120" @.***> 写道:

这个比赛说明写了,只有注册成为参赛者(审核通过需要1-2天),才能获取数据,你们可以看看。 我这边不太确定我能不能传数据到网盘(不知道会不会侵权啥的),这个数据集本身也很大,有20多G,我这两天可以先传一下,到时候我可能只敢开放有限的时间供你们下载,就担心侵权啥的。

在 2022-09-30 11:44:18,"JackWhite-rwx" @.***> 写道:

您可以在“https://competitions.codalab.org/competitions/22225”链接的参与选项卡中获取它。

这个比赛已经结束了,这个链接也无法再下载数据了,请问有保存当时训练和验证的数据吗?如果有的话,可否通过网盘链接分享一下?非常感谢您杰出的工作!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

Deep-imagelab avatar Oct 01 '22 08:10 Deep-imagelab

这个比赛说明写了,只有注册成为参赛者(审核通过需要1-2天),才能获取数据,你们可以看看。 我这边不太确定我能不能传数据到网盘(不知道会不会侵权啥的),这个数据集本身也很大,有20多G,我这两天可以先传一下,到时候我可能只敢开放有限的时间供你们下载,就担心侵权啥的。

在 2022-09-30 11:44:18,"JackWhite-rwx" @.***> 写道:

您可以在“https://competitions.codalab.org/competitions/22225”链接的参与选项卡中获取它。

这个比赛已经结束了,这个链接也无法再下载数据了,请问有保存当时训练和验证的数据吗?如果有的话,可否通过网盘链接分享一下?非常感谢您杰出的工作!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

Deep-imagelab avatar Oct 11 '22 07:10 Deep-imagelab

好的,谢谢。

------------------ 原始邮件 ------------------ 发件人: "Deep-imagelab/AWAN" @.>; 发送时间: 2022年10月1日(星期六) 下午4:43 @.>; @.@.>; 主题: Re: [Deep-imagelab/AWAN] Train data (#2)

我把数据传好了,有效期是30天,你们可以尽快下载下。

链接:https://pan.baidu.com/s/1BjvPWlXbjGxmuUZmH5tnGw

提取码:2022

在 2022-09-30 15:02:45,"img_wcx1120" @.***> 写道:

这个比赛说明写了,只有注册成为参赛者(审核通过需要1-2天),才能获取数据,你们可以看看。 我这边不太确定我能不能传数据到网盘(不知道会不会侵权啥的),这个数据集本身也很大,有20多G,我这两天可以先传一下,到时候我可能只敢开放有限的时间供你们下载,就担心侵权啥的。

在 2022-09-30 11:44:18,"JackWhite-rwx" @.***> 写道:

您可以在“https://competitions.codalab.org/competitions/22225”链接的参与选项卡中获取它。

这个比赛已经结束了,这个链接也无法再下载数据了,请问有保存当时训练和验证的数据吗?如果有的话,可否通过网盘链接分享一下?非常感谢您杰出的工作!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.>

msn321 avatar Oct 11 '22 07:10 msn321

@msn321 可以分享下百度链接吗?百度链接失效了:blush:

newshu avatar Nov 24 '22 02:11 newshu

https://pan.baidu.com/s/1BjvPWlXbjGxmuUZmH5tnGw

链接: https://pan.baidu.com/s/1GMxU6ojEi98n134JYdnjMg 提取码: hkm3

Deep-imagelab avatar Nov 28 '22 02:11 Deep-imagelab

@Deep-imagelab 感谢大佬。再冒昧问一下,我看见你在MST++下评论无法跑出作者论文公布的结果,我在2080TI上也没有跑出额,请问后续大佬在版本更改后跑出作者论文结果了吗?

newshu avatar Nov 28 '22 06:11 newshu

我后来训练网络的时候没有直接使用MST++的源代码,而是仅用了MST++的网络结构,放在了我的代码框架下训练的。 你也可以试着改一下MST++的源代码,我记得好像是因为加载数据的那里会导致电脑内存不够用

------------------ 原始邮件 ------------------ 发件人: "Deep-imagelab/AWAN" @.>; 发送时间: 2022年11月28日(星期一) 下午2:25 @.>; @.@.>; 主题: Re: [Deep-imagelab/AWAN] Train data (#2)

@Deep-imagelab 感谢大佬。再冒昧问一下,我看见你在MST++下评论无法跑出作者论文公布的结果,我在2080TI上也没有跑出额,请问后续大佬在版本更改后跑出作者论文结果了吗?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

msn321 avatar Nov 28 '22 09:11 msn321

@Deep-imagelab 感谢大佬。再冒昧问一下,我看见你在MST++下评论无法跑出作者论文公布的结果,我在2080TI上也没有跑出额,请问后续大佬在版本更改后跑出作者论文结果了吗?

我跑了几次也没跑出来,总是差一点点,可能是随机性的影响吧

Deep-imagelab avatar Nov 29 '22 12:11 Deep-imagelab

我在3090上使用pytorch1.9.0_cu11.1跑的结果比论文里的高

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2022年11月29日(星期二) 晚上8:26 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [Deep-imagelab/AWAN] Train data (#2)

@Deep-imagelab 感谢大佬。再冒昧问一下,我看见你在MST++下评论无法跑出作者论文公布的结果,我在2080TI上也没有跑出额,请问后续大佬在版本更改后跑出作者论文结果了吗?

我跑了几次也没跑出来,总是差一点点,可能是随机性的影响吧

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

JackWhite-rwx avatar Nov 30 '22 09:11 JackWhite-rwx

好的,非常感谢

------------------ 原始邮件 ------------------ 发件人: "Deep-imagelab/AWAN" @.>; 发送时间: 2022年10月1日(星期六) 下午4:43 @.>; @.@.>; 主题: Re: [Deep-imagelab/AWAN] Train data (#2)

我把数据传好了,有效期是30天,你们可以尽快下载下。

链接:https://pan.baidu.com/s/1BjvPWlXbjGxmuUZmH5tnGw

提取码:2022

在 2022-09-30 15:02:45,"img_wcx1120" @.***> 写道:

这个比赛说明写了,只有注册成为参赛者(审核通过需要1-2天),才能获取数据,你们可以看看。 我这边不太确定我能不能传数据到网盘(不知道会不会侵权啥的),这个数据集本身也很大,有20多G,我这两天可以先传一下,到时候我可能只敢开放有限的时间供你们下载,就担心侵权啥的。

在 2022-09-30 11:44:18,"JackWhite-rwx" @.***> 写道:

您可以在“https://competitions.codalab.org/competitions/22225”链接的参与选项卡中获取它。

这个比赛已经结束了,这个链接也无法再下载数据了,请问有保存当时训练和验证的数据吗?如果有的话,可否通过网盘链接分享一下?非常感谢您杰出的工作!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.> — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.>

JackWhite-rwx avatar Feb 11 '23 09:02 JackWhite-rwx