Home>

keras is used in Windows10 + Anaconda environment.
I recently learned that tqdm is useful and started using it.
Apparently, when tqdm is imported in my environment, the log during learning of keras is broken.
Even if you don't actually use tqdm in your code, it will collapse.
In my environment, the reproducibility is 100% so far, so tqdm is definitely bad.
In terms of behavior, it seems that "\ r" does not seem to work well inside.
Reinstalling Anaconda, creating a new virtual environment, installing previous versions of tqdm,
I tried to install via tqdm via conda and pip, but it has not changed.
If anyone knows how to solve it, thank you.

Error message

MNIST tutorial code execution example

  • tqdm imported
Train on 60000 samples, validate on 10000 samples
Epoch 1/20
60000/60000 [==============================]-ETA: 14:10-loss: 2.3458-acc: 0.07- ETA: 1:17-loss: 1.3900-acc: 0.5391-ETA: 40s-loss: 1.0298-acc: 0.658-ETA: 27s-loss: 0.8557-acc: 0.71-ETA: 20s-loss: 0.7458-acc: 0.75- ETA: 16s-loss: 0.6699-acc: 0.78-ETA: 13s-loss: 0.6096-acc: 0.80-ETA: 11s-loss: 0.5708-acc: 0.81-ETA: 9s-loss: 0.5340-acc: 0.8280-ETA: 8s-loss: 0.5024-acc: 0.839-ETA: 7s-loss: 0.4833-acc: 0.846-ETA: 6s-loss: 0.4624-acc: 0.853-ETA: 6s-loss: 0.4440-acc: 0.859-ETA: 5s- loss: 0.4305-acc: 0.865-ETA: 5s-loss: 0.4201-acc: 0.868-ETA: 5s-loss: 0.4065-acc: 0.872-ETA: 4s-loss: 0.3941-acc: 0.875-ETA: 4s-loss: 0.3851-acc: 0.878-ETA: 4s-loss: 0.3758-acc: 0.882-ETA: 3s-loss: 0.3670-acc: 0.884-ETA: 3s-loss: 0.3577-acc: 0.888-ETA: 3s-loss: 0.3499- acc: 0.890-ETA: 3s-loss: 0.3418-acc: 0.893-ETA: 2s-loss: 0.3340-acc: 0.895-ETA: 2s-loss: 0.3281-acc: 0.89 7-ETA: 2s-loss: 0.3223-acc: 0.899-ETA: 2s-loss: 0.3187-acc: 0.900-ETA: 2s-loss: 0.3132-acc: 0.902-ETA: 1s-loss: 0.3088-acc: 0.903- ETA: 1s-loss: 0.3035-acc: 0.905-ETA: 1s-loss: 0.2979-acc: 0.907-ETA: 1s-loss: 0.2930-acc: 0.908-ETA: 1s-loss: 0.2883-acc: 0.910-ETA: 1s-loss: 0.2845-acc: 0.911-ETA: 1s-loss: 0.2805-acc: 0.912-ETA: 1s-loss: 0.2771-acc: 0.913-ETA: 0s-loss: 0.2731-acc: 0.914-ETA: 0s- loss: 0.2696-acc: 0.915-ETA: 0s-loss: 0.2659-acc: 0.916-ETA: 0s-loss: 0.2635-acc: 0.917-ETA: 0s-loss: 0.2606-acc: 0.918-ETA: 0s-loss: 0.2574-acc: 0.919-ETA: 0s-loss: 0.2544-acc: 0.920-ETA: 0s-loss: 0.2511-acc: 0.921-ETA: 0s-loss: 0.2493-acc: 0.922-ETA: 0s-loss: 0.2479- acc: 0.922-ETA: 0s-loss: 0.2455-acc: 0.923-4s 73us/step-loss: 0.2454-acc: 0.9235-val_loss: 0.1531-val_acc: 0.9523
Epoch 2/20
60000/60000 [==============================]-ETA: 4s-loss: 0.1293-acc: 0.953-ETA: 3s-loss: 0.1000-acc: 0.969-ETA: 2s-loss: 0.1068-acc: 0.967-ETA: 2s-loss: 0.1030-acc: 0.969-ETA: 2s-loss: 0.1066-acc: 0.968-ETA: 2s- loss: 0.1034-acc: 0.970-ETA: 2s-loss: 0.1036-acc: 0.969-ETA: 2s-loss: 0.1065-acc: 0.969-ETA: 2s-loss: 0.1084-acc: 0.967-ETA: 2s-loss: 0.1123-acc: 0.966-ETA: 2s-loss: 0.1091-acc: 0.967-ETA: 2s-loss: 0.1086-acc: 0.967-ETA: 2s-loss: 0.1078-acc: 0.968-ETA: 2s-loss: 0.1057- acc: 0.968-ETA: 2s-loss: 0.1054-acc: 0.968-ETA: 1s-loss: 0.1083-acc: 0.968-ETA: 1s-loss: 0.1077-acc: 0.968-ETA: 1s-loss: 0.1101-acc: 0.967-ETA: 1s-loss: 0.1099-acc: 0.967-ETA: 1s-loss: 0.1097-acc: 0.967-ETA: 1s-loss: 0.1092-acc: 0.967-ETA: 1s-loss: 0.1094-acc: 0.967- ETA: 1s-loss: 0.1090-acc: 0.967-ETA: 1s-loss: 0.1085-acc: 0.968-ETA: 1s-loss: 0.1084-acc: 0.968-ETA : 1s-loss: 0.1083-acc: 0.968-ETA: 1s-loss: 0.1085-acc: 0.967-ETA: 1s-loss: 0.1079-acc: 0.967-ETA: 1s-loss: 0.1080-acc: 0.968-ETA: 1s -loss: 0.1077-acc: 0.968-ETA: 1s-loss: 0.1075-acc: 0.968-ETA: 1s-loss: 0.1071-acc: 0.968-ETA: 1s-loss: 0.1067-acc: 0.968-ETA: 1s-loss : 0.1069-acc: 0.968-ETA: 1s-loss: 0.1067-acc: 0.968-ETA: 0s-loss: 0.1065-acc: 0.968-ETA: 0s-loss: 0.1066-acc: 0.968-ETA: 0s-loss: 0.1063 -acc: 0.968-ETA: 0s-loss: 0.1063-acc: 0.968-ETA: 0s-loss: 0.1056-acc: 0.968-ETA: 0s-loss: 0.1051-acc: 0.969-ETA: 0s-loss: 0.1051-acc : 0.969-ETA: 0s-loss: 0.1045-acc: 0.969-ETA: 0s-loss: 0.1044-acc: 0.969-ETA: 0s-loss: 0.1044-acc: 0.969-ETA: 0s-loss: 0.1039-acc: 0.969 -ETA: 0s-loss: 0.1040-acc: 0.969-ETA: 0s-loss: 0.1044-acc: 0.969-ETA: 0s-loss: 0.1037-acc: 0.969-ETA: 0s-loss: 0.1042-acc: 0.969-ETA : 0s-loss: 0.1038-acc: 0.969-ETA: 0s-loss: 0.1035-acc: 0.969-ETA: 0s-loss: 0.1032-acc: 0.969-ETA: 0s-loss: 0.1035-acc: 0.969-ETA: 0s-loss: 0.1035-acc: 0.969-3s 50us/step-loss: 0.1035 -acc: 0.9694-val_loss: 0.0872-val_acc: 0.9731
  • No tqdm import
Train on 60000 samples, validate on 10000 samples
Epoch 1/2060000/60000 [=======================]-4s 70us/step-loss: 0.2448-acc: 0.9247-val_loss : 0.1020-val_acc: 0.9681
Epoch 2/20
60000/60000 [=======================]-2s 41us/step-loss: 0.0998-acc: 0.9690-val_loss : 0.0814-val_acc: 0.9742
Supplemental information (FW/tool version etc.)

Windows10
Anaconda 1.9.7
python 3.6.8
tensorflow-gpu 1.14.0
keras 2.2.4
tqdm 4.25.0