Linear Regression example doesn't work.

[Esther2013]

Linear Regression example doesn't work.

create_op: NoOp 'GradientDescent', inputs: empty, control_inputs: GradientDescent/update_weight/ApplyGradientDescent, GradientDescent/update_bias/ApplyGradientDescent, outputs: empty
create_op: NoOp 'init', inputs: empty, control_inputs: weight/Assign, bias/Assign, outputs: empty
Epoch: 50 cost=NaN W=NaN b=NaN
Epoch: 100 cost=NaN W=NaN b=NaN

[Oceania2018]

Got some clue. There are only 95 nodes, ideally there should be 121 nodes. Have to figure out what is missing.

[Oceania2018]

Added all operations but still not running, try to export meta graph and import.

[Oceania2018]

It's working now. Check the code here:

2/22/2019 5:46:00 AM Starting LinearRegression
2019-02-21 23:46:01.084687: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
Epoch: 50 cost=0.1932824 W=0.440269 b=-0.5702394
Epoch: 100 cost=0.1798451 W=0.4289288 b=-0.4886577
Epoch: 150 cost=0.1679595 W=0.4182628 b=-0.4119281
Epoch: 200 cost=0.1574465 W=0.4082314 b=-0.3397623
Epoch: 250 cost=0.1481478 W=0.3987965 b=-0.2718883
Epoch: 300 cost=0.1399229 W=0.3899226 b=-0.208051
Epoch: 350 cost=0.1326481 W=0.3815766 b=-0.1480103
Epoch: 400 cost=0.1262135 W=0.373727 b=-0.09154031
Epoch: 450 cost=0.1205223 W=0.3663442 b=-0.038429
Epoch: 500 cost=0.1154885 W=0.3594005 b=0.01152356
Epoch: 550 cost=0.1110362 W=0.3528697 b=0.05850525
Epoch: 600 cost=0.1070983 W=0.3467275 b=0.1026927
Epoch: 650 cost=0.1036154 W=0.3409505 b=0.1442519
Epoch: 700 cost=0.100535 W=0.335517 b=0.1833397
Epoch: 750 cost=0.09781052 W=0.3304067 b=0.2201028
Epoch: 800 cost=0.09540092 W=0.3256003 b=0.2546793
Epoch: 850 cost=0.09326982 W=0.3210799 b=0.2871991
Epoch: 900 cost=0.09138502 W=0.3168282 b=0.3177855
Epoch: 950 cost=0.08971816 W=0.3128294 b=0.3465521
Epoch: 1000 cost=0.08824395 W=0.3090685 b=0.3736083
Optimization Finished!
Training cost=0.08824395 W=0.3090685 b=0.3736083
Testing... (Mean square loss Comparison)
Testing cost=0.07984468
Absolute mean square loss difference: 0.008399263
2/22/2019 5:46:04 AM Completed LinearRegression