Which seeds have to be set where to realize 100% reproducibility of training results in tensorflow?

by

The best solution which works as of today with GPU is to install tensorflow-determinism with the following:

pip install tensorflow-determinism

Then include the following code to your code

import tensorflow as tf
import os
os.environ['TF_DETERMINISTIC_OPS'] = '1'

source: https://github.com/NVIDIA/tensorflow-determinism

More Related Contents:

  1. Reproducible results in Tensorflow with tf.set_random_seed
  2. What are logits? What is the difference between softmax and softmax_cross_entropy_with_logits?
  3. TensorFlow Only running on 1/32 of the Training data provided [duplicate]
  4. How to get current available GPUs in tensorflow?
  5. How to export Keras .h5 to tensorflow .pb?
  6. How to choose cross-entropy loss in TensorFlow?
  7. Keras + Tensorflow and Multiprocessing in Python
  8. Is it still necessary to install CUDA before using the conda tensorflow-gpu package?
  9. Error running basic tensorflow example
  10. Is Tensorflow compatible with a Windows workflow?
  11. Keras Sequential model with multiple inputs
  12. ValueError at /image/ Tensor Tensor(“activation_5/Softmax:0”, shape=(?, 4), dtype=float32) is not an element of this graph
  13. Streaming large training and test files into Tensorflow’s DNNClassifier
  14. TensorFlow operator overloading
  15. How to get other metrics in Tensorflow 2.0 (not only accuracy)?
  16. Clearing Tensorflow GPU memory after model execution
  17. Removing then Inserting a New Middle Layer in a Keras Model
  18. Why can’t I get reproducible results in Keras even though I set the random seeds?
  19. NotImplementedError: Cannot convert a symbolic Tensor (lstm_2/strided_slice:0) to a numpy array. T
  20. How convert a jpeg image into json file in Google machine learning
  21. AttributeError: module ‘tensorflow.python.keras.utils.generic_utils’ has no attribute ‘populate_dict_with_module_objects’
  22. How to get stable results with TensorFlow, setting random seed
  23. “freeze” some variables/scopes in tensorflow: stop_gradient vs passing variables to minimize
  24. How can I visualize the weights(variables) in cnn in Tensorflow?
  25. Tensor is not an element of this graph
  26. overcome Graphdef cannot be larger than 2GB in tensorflow
  27. Training a simple model in Tensorflow GPU slower than CPU
  28. Tensorflow One Hot Encoder?
  29. cannot import name ‘_registerMatType’ from ‘cv2.cv2’
  30. How to do Xavier initialization on TensorFlow

Leave a Comment