Error: Batch normalization gradient requires mean and offset to have equal ranks.
See original GitHub issueHello Everyone,
[pix2pix
][pix2pix.js
][tensorflow.js
]
I’m trying to set up online environment for my new learned pix2pix model. I have exported the .pict model with the guides kindly provided by @yining1023 & Dongphil Yoo.
TensorFlow.js version
https://cdn.jsdelivr.net/npm/@tensorflow/tfjs
Browser version
Google Chrome Version 77.0.3865.90 (Build officiel) (64 bits) FIrefox quantum 69…0.1
Describe the problem or feature request
After .pict Model is loaded Anything done with tools (Eraser/clear/pencil) produces error as following in Console
tfjs:2 Uncaught (in promise) Error: Batch normalization gradient requires mean and offset to have equal ranks. at f (tfjs:2) at As (tfjs:2) at Object.batchNorm (tfjs:2) at batchnorm (pix2pix.js:34) at pix2pix.js:64 at tfjs:2 at t.scopedRun (tfjs:2) at t.tidy (tfjs:2) at Object._e [as tidy] (tfjs:2) at Pix2pix.transfer (pix2pix.js:47)
Code to reproduce the bug / link to feature request
http://guillaumedamry.com/p2pampa/
Any hint on this (programming or orienting request on github) would be much appreciated because I am completly stuck.
I have compiled with python3 [pix2pix
][pix2pix.js
][tensorflow.js
]
Issue Analytics
- State:
- Created 4 years ago
- Comments:6
Top GitHub Comments
Just to be clear #2114 is not meant to be a fix to this problem. It’s just relevant to the problem and shows how a user may in principle workaround it. The fix will come in a separate PR.
Closing this due to lack of activity, feel to reopen. Thank you