Closed apozas closed 7 years ago
Wow, 97% accuracy on the real-world data with data augmentation, and with only a few minutes of training! This is awesome. On the other hand, adding the h5 file was not a great idea: it is a generated file, which by principle of version control systems, should not be there, plus it is a 90MByte binary that now will be there forever (deleting is not an option since git will remember it anyway).
@peterwittek - i'm not sure if it can be used after the fact, but maybe check out - https://git-lfs.github.com/
I was trying bfg, but did not get very far.
Okay, I think it is resolved now. Both the binary blob and its history vanished. @apozas, please delete all local clones as well as your fork. Then fork it again and clone. The history of your current fork is inconsistent with the altered history of the fixed-up upstream version.
OK, steps performed, everything looks well. My apologies for this mess, it will not happen again.
2017-03-21 14:16 GMT+01:00 Peter Wittek notifications@github.com:
Okay, I think it is resolved now. Both the binary blob and its history vanished. @apozas https://github.com/apozas, please delete all local clones as well as your fork. Then fork it again and clone. The history of your current fork is inconsistent with the altered history of the fixed-up upstream version.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/peterwittek/qml-rg/pull/24#issuecomment-288074609, or mute the thread https://github.com/notifications/unsubscribe-auth/AYmgpqkWmNcxcrnc9o3D-kFrl8JcoIHGks5rn82sgaJpZM4Milqs .
Code (APS_Cracker.ipynb) + trained weights (APS_Cracker_weights.h5)