linnanwang / alphax-nasbench101 Goto Github PK

Awesome work! I trained net_predictor with default settings (Adding validation_split=.2). The val_correlation can achieve 0.819, which is much bigger than the number (0.784) in the paper. Is there anything wrong? Thanks.

Hi, May I ask what tools you used to plot this figure? Thank you very much! I really appreciate this!

Hello, as you have recommended, I studied your new paper and was amazed by the result, and the idea was quite enlightening, however, there are a few things that still confuses me:

1. Does Xbar in node.Xbar stand for the mean accuracy of architectures attached to that node
2. I am very curious about the way you encode the architectures, does it affect the performance a lot?
   Thank you very much and looking forward to your reply!

Thank you for the code! It would be great if you can share the RNN predictor code as well, because I'd like to try both and see the result. Great thanks!

Cannot correctly download the NASBench dataset with the provided link.

Thank you for the great idea in this work!
I would like to ask about the difference of the search space based on NASNet between AlphaX and NASNet. It seems that in each block of a cell, either input hidden state can be passed through up to two layers with different operations. In my understanding, it is not the same as NASNet, which only exists one operation for either input hidden state. Is the setting made up purposely?
Please let me know if I misunderstood.
Thanks again.