Darts: Differentiable architecture search for convolutional, recurrent networks

Re:Darts: Differentiable architecture search for convolution... by NhanH on Friday July 13, 2018 @01:14PM (#17524791)

Can you link the paper of Zoph,Le that did the “1000 GPUs” thing? Thanks! I tried searching for it but apparently there was quite a few papers on parameter search.

[ Parent | on HN ]

Re:Darts: Differentiable architecture search for convolution... by kripke on Friday July 13, 2018 @02:55PM (#17525701)

Not GP, but you are probably looking for the "Neural Architecture Search" series [1] [2] [3]. First one uses something like 1k GPUs for a month, next one is a bit more reasonable, and the last one actually has a reasonable training time, and has comparable performance to DARTS (see ENAS in comparison tables).

1: https://arxiv.org/abs/1611.01578 2: https://arxiv.org/abs/1707.07012 3: https://arxiv.org/abs/1802.03268

[ Parent | on HN ]