Wenhao Wang

@or-toledano @bryant1410 @jongwook @jiahuei Looks forward to your reply. Thanks in advance.

@kalyo-zjl I have finished a paper to address the unstable problem and hope it will work: https://arxiv.org/abs/2002.09053

Maybe there is something wrong with your Pytorch version.

Maybe I will write a paper to discuss this problem more specifically. Please wait for it’s release. 发自我的 iPad ------------------ Original ------------------ From: LIGang

@LIszu the paper is available: https://arxiv.org/abs/2002.09053

"rerank_dist = compute_jaccard_distance(target_features, k1=args.k1, k2=args.k2, search_option=3, use_float16=True" It's a pity that it does not work. A server with 384G memory is still facing the pointed error when the training set...

Thanks for your reply～ 发自我的iPhone ------------------ Original ------------------ From: Yixiao Ge

Hahaha, this issue is given by myself.

差这么多的话就不是哪里可以改进的事情了；是出现问题了 发自我的iPhone ------------------ 原始邮件 ------------------ 发件人: yaru-w ***@***.***> 发送时间: 2021年6月9日 22:08 收件人: WangWenhao0716/Adapted-Center-and-Scale-Prediction ***@***.***> 抄送: Subscribed ***@***.***> 主题: 回复：[WangWenhao0716/Adapted-Center-and-Scale-Prediction] 你好，我想问一下我用您给的源代码进行了训练，但用自己重新训练出来的模型ACSP_150.pth.tea进行test，得到的结果却很差，这是为什么呢 (#18)

我的这个必须要2卡和我给那个batch size跑；8卡结果不一样是正常的 差1-2%？ 发自我的iPhone ------------------ 原始邮件 ------------------ 发件人: yaru-w ***@***.***> 发送时间: 2021年6月11日 22:38 收件人: WangWenhao0716/Adapted-Center-and-Scale-Prediction ***@***.***> 抄送: Wenhao Wang ***@***.***>, Comment ***@***.***> 主题: 回复：[WangWenhao0716/Adapted-Center-and-Scale-Prediction] 你好，我想问一下我用您给的源代码进行了训练，但用自己重新训练出来的模型ACSP_150.pth.tea进行test，得到的结果却很差，这是为什么呢 (#18) 那可能什么问题呢 我又跑了一遍 8块gpu 每块2个...