ninjia error

[littlestar231]

Hi! I want to train your model use my own dataset. Under the environment you used: pytorch 1.1.0, python 3.7, cuda 10.0. But unfortunately, it shows the error "ninja: build stopped: subcommand failed.\n" . Do you meet the same thing? Thanks a lot for your help.

[feinanshan]

Hi! Can you provide more information? Like at which step this error happens and other printouts on the screen.

[littlestar231]

Dear sir:      Thanks for your reply! I met the mistake with ninja(1.9.0) which is a package installed with pytorch 1.1.0. I run the code on the server at school. But unfortunately, there is wrong with it. I still have no idea after searching for some material. Do you meet the same problem? Sincerely for your help!

    Best wishes!

------------------ 原始邮件 ------------------ 发件人: "feinanshan/TDNet" <notifications@github.com>; 发送时间: 2020年7月16日(星期四) 晚上9:54 收件人: "feinanshan/TDNet"<TDNet@noreply.github.com>; 抄送: "飞虎联盟"<1160632674@qq.com>;"Author"<author@noreply.github.com>; 主题: Re: [feinanshan/TDNet] ninjia error (#5)

Hi! Can you provide more information? Like at which step this error happens and other printouts on the screen.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[feinanshan]

Hi, I didn't meet the error as you mentioned. Maybe try with Anaconda to prepare the environment for running?

[zhixuanli]

This "ninja error" is the problem of pytorch-encoding, and I'm also using the same configuration with @feinanshan but facing the same error. Still working to solve it.

[zhixuanli]

Finally, I created a new Anaconda env and reinstall everything. No error now. @littlestar231