Could you please tell me something about your strategy

[HONGJUNRL]

In the traditional darts methods, weights and architecture are optimized step by step, but in your paper, weights and architecture are optimized by epoch, is it mean that in this epoch,the model optimizes weights and in the next epoch, the model optimizes the architecture. I am quite confused.

[JaminFong]

alternate the optimization process of weights and architecture by epoch

It means that we split one search epoch into two parts. We first optimize the operation weights on the training dataset and after that, we optimize the architecture on the validation dataset.

[HONGJUNRL]

Thanks for your reply 发自我的iPhone

…

------------------ Original ------------------ From: Jamin Fong <notifications@github.com> Date: Sun,Jun 30,2019 4:19 PM To: JaminFong/DenseNAS <DenseNAS@noreply.github.com> Cc: HONGJUNRL <1929810084@qq.com>, Author <author@noreply.github.com> Subject: Re: [JaminFong/DenseNAS] Could you please tell me something about your strategy (#2) alternate the optimization process of weights and architecture by epoch It means that we split one search epoch into two parts. We first optimize the operation weights on the training dataset and after that, we optimize the architecture on the validation dataset. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.