log

[cheun726]

hi ,Do you have a log of your training?I'd like to refer to it,thanks

[HarukiYqM]

Hi, unfortunately since I switched my institution this year, I no longer have access to the training logs stored on the server. But if you follow the standard training specification, you should be able to reproduce.

I remember it achieves relatively good performance on benchmarks (38.2x on set5 x2) after training 200 epochs and this may be a good time point to check the performance.

[cheun726]

unfortunately,i don't have GPUS,i can not reproduce.

…

------------------ 原始邮件 ------------------ 发件人: ***@***.***>; 发送时间: 2021年9月8日(星期三) 晚上9:00 收件人: ***@***.***>; 抄送: ***@***.***>; ***@***.***>; 主题: Re: [HarukiYqM/Non-Local-Sparse-Attention] log (#11) Hi, unfortunately since I switched my institution this year, I no longer have access to the training logs stored on the server. But if follow the standard training specification, you should be able to reproduce, and achieve relatively good performance on benchmarks even only train 200 epochs. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[cheun726]

hi  Why did this happen when I tested the model I trained myself

…

------------------ 原始邮件 ------------------ 发件人: "HarukiYqM/Non-Local-Sparse-Attention" ***@***.***>; 发送时间: 2021年9月8日(星期三) 晚上9:00 ***@***.***>; ***@***.******@***.***>; 主题: Re: [HarukiYqM/Non-Local-Sparse-Attention] log (#11) Hi, unfortunately since I switched my institution this year, I no longer have access to the training logs stored on the server. But if follow the standard training specification, you should be able to reproduce, and achieve relatively good performance on benchmarks even only train 200 epochs. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.