How to get stable results with TensorFlow, setting random seed

by

Setting the current TensorFlow random seed affects the current default graph only. Since you are creating a new graph for your training and setting it as default (with g.as_default():), you must set the random seed within the scope of that with block.

For example, your loop should look like the following:

for i in range(3):
  g = tf.Graph()
  with g.as_default():
    tf.set_random_seed(1)
    accuracy_result, average_error = network.train_network(
        parameters, inputHeight, inputWidth, inputChannels, outputClasses)

Note that this will use the same random seed for each iteration of the outer for loop. If you want to use a different—but still deterministic—seed in each iteration, you can use tf.set_random_seed(i + 1).

More Related Contents:

  1. Convert a tensor to numpy array in Tensorflow?
  2. Using a pre-trained word embedding (word2vec or Glove) in TensorFlow
  3. AttributeError: ‘Tensor’ object has no attribute ‘numpy’
  4. TypeError: __array__() takes 1 positional argument but 2 were given (Image classification Keras)
  5. Conditional assignment of tensor values in TensorFlow
  6. TensorFlow ValueError: Cannot feed value of shape (64, 64, 3) for Tensor u’Placeholder:0′, which has shape ‘(?, 64, 64, 3)’
  7. NotImplementedError: Cannot convert a symbolic Tensor (lstm_2/strided_slice:0) to a numpy array. T
  8. Load numpy array in google-cloud-ml job
  9. Why is numpy native on M1 Max greatly slower than on old Intel i5?
  10. Taking subarrays from numpy array with given stride/stepsize
  11. ‘and’ (boolean) vs ‘&’ (bitwise) – Why difference in behavior with lists vs numpy arrays?
  12. Moving average or running mean
  13. Compiling numpy with OpenBLAS integration
  14. Python/NumPy first occurrence of subarray
  15. What is the internal precision of numpy.float128?
  16. Use a.any() or a.all()
  17. Restart cumsum and get index if cumsum more than value
  18. ValueError: numpy.dtype has the wrong size, try recompiling
  19. Implement MATLAB’s im2col ‘sliding’ in Python
  20. subsampling every nth entry in a numpy array
  21. Find matching rows in 2 dimensional numpy array
  22. Python’s sum vs. NumPy’s numpy.sum
  23. Can Keras with Tensorflow backend be forced to use CPU or GPU at will?
  24. Pandas lookup from one of multiple columns, based on value
  25. When should I use hstack/vstack vs append vs concatenate vs column_stack?
  26. Type hinting / annotation (PEP 484) for numpy.ndarray
  27. Pass percentiles to pandas agg function
  28. remove zero lines 2-D numpy array
  29. 2D and 3D Scatter Histograms from arrays in Python
  30. How to prefetch data using a custom python function in tensorflow

Leave a Comment