How to get reproducible results in keras

by

You can find the answer at the Keras docs: https://keras.io/getting-started/faq/#how-can-i-obtain-reproducible-results-using-keras-during-development.

In short, to be absolutely sure that you will get reproducible results with your python script on one computer’s/laptop’s CPU then you will have to do the following:

  1. Set the PYTHONHASHSEED environment variable at a fixed value
  2. Set the python built-in pseudo-random generator at a fixed value
  3. Set the numpy pseudo-random generator at a fixed value
  4. Set the tensorflow pseudo-random generator at a fixed value
  5. Configure a new global tensorflow session

Following the Keras link at the top, the source code I am using is the following:

# Seed value
# Apparently you may use different seed values at each stage
seed_value= 0

# 1. Set the `PYTHONHASHSEED` environment variable at a fixed value
import os
os.environ['PYTHONHASHSEED']=str(seed_value)

# 2. Set the `python` built-in pseudo-random generator at a fixed value
import random
random.seed(seed_value)

# 3. Set the `numpy` pseudo-random generator at a fixed value
import numpy as np
np.random.seed(seed_value)

# 4. Set the `tensorflow` pseudo-random generator at a fixed value
import tensorflow as tf
tf.random.set_seed(seed_value)
# for later versions: 
# tf.compat.v1.set_random_seed(seed_value)

# 5. Configure a new global `tensorflow` session
from keras import backend as K
session_conf = tf.ConfigProto(intra_op_parallelism_threads=1, inter_op_parallelism_threads=1)
sess = tf.Session(graph=tf.get_default_graph(), config=session_conf)
K.set_session(sess)
# for later versions:
# session_conf = tf.compat.v1.ConfigProto(intra_op_parallelism_threads=1, inter_op_parallelism_threads=1)
# sess = tf.compat.v1.Session(graph=tf.compat.v1.get_default_graph(), config=session_conf)
# tf.compat.v1.keras.backend.set_session(sess)

It is needless to say that you do not have to to specify any seed or random_state at the numpy, scikit-learn or tensorflow/keras functions that you are using in your python script exactly because with the source code above we set globally their pseudo-random generators at a fixed value.

More Related Contents:

  1. Deep-Learning Nan loss reasons
  2. Saving to hdf5 is very slow (Python freezing)
  3. NaN loss when training regression network
  4. Getting gradient of model output w.r.t weights using Keras
  5. TypeError: __array__() takes 1 positional argument but 2 were given (Image classification Keras)
  6. How to compute Receiving Operating Characteristic (ROC) and AUC in keras?
  7. Keras import error Nadam
  8. Keras: “RuntimeError: Failed to import pydot.” after installing graphviz and pydot
  9. memory issues when transforming np.array using to_categorical
  10. Keras accuracy does not change
  11. How to fix ‘Object arrays cannot be loaded when allow_pickle=False’ for imdb.load_data() function?
  12. Keras, how do I predict after I trained a model?
  13. How to shade region under the curve in matplotlib
  14. How to resample a dataframe with different functions applied to each column?
  15. Randomly select from numpy array
  16. Extract csv file specific columns to list in Python
  17. Resample time series in pandas to a weekly interval
  18. How to convert an image from np.uint16 to np.uint8?
  19. Python import error: cannot import name ‘six’ from ‘sklearn.externals’
  20. Consecutive, Overlapping Subsets of Array (NumPy, Python)
  21. How to use NumPy array with ctypes?
  22. Can’t import cv2; “DLL load failed”
  23. A numpy array unexpectedly changes when changing another one despite being separate
  24. How to use numpy to get the cumulative count by unique values in linear time?
  25. numpy.sum() giving strange results on large arrays
  26. Computing cross-correlation function?
  27. Annotate Time Series plot
  28. Understanding scipy deconvolve
  29. Numpy `ValueError: operands could not be broadcast together with shape …`
  30. Avoid overflow when adding numpy arrays

Leave a Comment