Merge pull request #28 from N3PDF/migrate_tf2

Migrate TF1 to TF2

.github/workflows/pythonapp.yml

@@ -7,9 +7,9 @@ jobs:
 
     runs-on: ubuntu-latest
     strategy:
-        max-parallel: 2
+        max-parallel: 3
         matrix:
-            python-version: [3.6, 3.7]
+            python-version: [3.6, 3.7, 3.8]
 
     steps:
     - uses: actions/checkout@v1
@@ -20,6 +20,7 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
+        pip install setuptools --upgrade
         pip install .
     - name: Lint with pylint
       run: |
@@ -31,5 +32,4 @@ jobs:
     - name: Test with pytest
       run: |
         pip install pytest
-        pip install tensorflow # Keras does not install any backend by default?
         pytest

.readthedocs.yml

@@ -6,4 +6,9 @@ sphinx:
 
 python:
   version: 3.7
-  pip_install: True
+  install:
+      - method: pip
+        path: .
+        extra_requirements:
+            - docs
+  system_packages: true

doc/source/conf.py

@@ -26,6 +26,7 @@
 
 # The full version, including alpha/beta/rc tags
 release = evolutionary_keras .__version__
+autodoc_mock_imports = ['tensorflow']
 
 
 # -- General configuration ---------------------------------------------------

examples/test.py

@@ -1,21 +1,25 @@
 """ Example implementation of evolutionary_keras
 """
 import sys
-import keras
-from keras.datasets import mnist
-from keras import backend as K
-from keras.layers import Dense, Input, Flatten
-
+import tensorflow as tf
+from tensorflow.keras.datasets import mnist
+from tensorflow.keras import backend as K
+from tensorflow.keras.layers import Dense, Input, Flatten
 from evolutionary_keras.models import EvolModel
 import evolutionary_keras.optimizers
 
+import logging
+logging.basicConfig(level = logging.INFO)
+logger = logging.getLogger()
+
+
 
 batch_size = 128
 num_classes = 10
 dense_size = 16
-epochs = 4000
+epochs = 40
 
-max_epochs = 40000
+max_epochs = 400
 
 # input image dimensions
 img_rows, img_cols = 28, 28
@@ -41,8 +45,8 @@
 print(x_test.shape[0], "test samples")
 
 # convert class vectors to binary class matrices
-y_train = keras.utils.to_categorical(y_train, num_classes)
-y_test = keras.utils.to_categorical(y_test, num_classes)
+y_train = tf.keras.utils.to_categorical(y_train, num_classes)
+y_test = tf.keras.utils.to_categorical(y_test, num_classes)
 
 inputs = Input(shape=(28, 28, 1))
 flatten = Flatten()(inputs)
@@ -56,7 +60,7 @@
     myopt = evolutionary_keras.optimizers.CMA(population_size=5, sigma_init=15)
     epochs = 1
 else:
-    myopt = evolutionary_keras.optimizers.NGA(population_size=20, sigma_init=15)
+    myopt = evolutionary_keras.optimizers.NGA(population_size=2, sigma_init=15)
 
 print(" > Compiling the model")
 model.compile(optimizer=myopt, loss="categorical_crossentropy", metrics=["accuracy"])
@@ -71,6 +75,7 @@
     verbose=1,
     # validation_data=(x_test, y_test)
 )
-score = model.evaluate(x=x_test, y=y_test, verbose=0)
-print("Test loss:", score[0])
-print("Test accuracy:", score[1])
+score = model.evaluate(x=x_test, y=y_test, return_dict=True, verbose=0)
+
+print("Test loss:", score['loss'])
+print("Test accuracy:", score['accuracy'])

setup.py

@@ -1,19 +1,44 @@
 from setuptools import setup, find_packages
-from os import path
+import os
+import re
 
-this_directory = path.abspath(path.dirname(__file__))
-with open(path.join(this_directory, "README.md"), encoding="utf-8") as f:
+
+requirements = ['numpy', 'cma']
+if os.environ.get('READTHEDOCS') != 'True':
+    requirements.append('tensorflow>2.1*')
+PACKAGE = 'evolutionary_keras'
+
+this_directory = os.path.abspath(os.path.dirname(__file__))
+with open(os.path.join(this_directory, "README.md"), encoding="utf-8") as f:
     long_description = f.read()
 
+def get_version():
+    """ Gets the version from the package's __init__ file
+    if there is some problem, let it happily fail """
+    VERSIONFILE = os.path.join('src', PACKAGE, '__init__.py')
+    initfile_lines = open(VERSIONFILE, 'rt').readlines()
+    VSRE = r"^__version__ = ['\"]([^'\"]*)['\"]"
+    for line in initfile_lines:
+        mo = re.search(VSRE, line, re.M)
+        if mo:
+            return mo.group(1)
+
 setup(
     name="evolutionary_keras",
-    version="1.0.1",
+    version=get_version(),
     author="S. Carrazza, J. Cruz-Martinez, Roy Stegeman",
     author_email="juan.cruz@mi.infn.it, roy.stegeman@mi.infn.it",
     url="https://github.com/N3PDF/evolutionary_keras",
     package_dir={"": "src"},
     packages=find_packages("src"),
-    install_requires=["numpy", "keras", "sphinx_rtd_theme", "recommonmark", "cma"],
+    zip_safe=False,
+    install_requires=requirements,
+    extras_require={
+        'docs' : [
+        'sphinx_rtd_theme',
+        'recommonmark',
+        ],
+    },
     python_requires=">=3.6",
     descriptions="An evolutionary algorithm implementation for Keras",
     long_description=long_description,

src/evolutionary_keras/__init__.py

@@ -1 +1 @@
-__version__ = "1.0.0"
+__version__ = "1.9"

src/evolutionary_keras/models.py

@@ -2,11 +2,10 @@
 
 import logging
 
-from keras.callbacks.callbacks import History
-from keras.models import Model
+from tensorflow.keras.callbacks import History
+from tensorflow.keras.models import Model
 
 import evolutionary_keras.optimizers as Evolutionary_Optimizers
-from evolutionary_keras.utilities import parse_eval
 
 log = logging.getLogger(__name__)
 
@@ -27,15 +26,16 @@ def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
         self.is_genetic = False
         self.opt_instance = None
-        self.history = History()
+        self.history_info = History()
 
     def parse_optimizer(self, optimizer):
         """ Checks whether the optimizer is genetic and creates and optimizer instance in case a
         string was given as input.
         """
         # Checks (if the optimizer input is a string) and whether it is in the 'optimizers'
         # dictionary
-        if isinstance(optimizer, str) and optimizer in optimizer_dict.keys():
+
+        if isinstance(optimizer, str) and optimizer.lower() in optimizer_dict.keys():
             opt = optimizer_dict.get(optimizer.lower())
             # And instanciate it with default values
             optimizer = opt()
@@ -50,6 +50,7 @@ def compile(self, optimizer="rmsprop", **kwargs):
         """ When the optimizer is genetic, compiles the model in keras setting an arbitrary
         keras supported optimizer """
         self.parse_optimizer(optimizer)
+        self.history_info.set_model(self)
         if self.is_genetic:
             super().compile(optimizer="rmsprop", **kwargs)
         else:
@@ -71,7 +72,7 @@ def perform_genetic_fit(
                 verbose, prints to log.info the loss per epoch
         """
         # Prepare the history for the initial epoch
-        self.history.on_train_begin()
+        self.history_info.on_train_begin()
         # Validation data is currently not being used!!
         if validation_data is not None:
             log.warning(
@@ -84,29 +85,24 @@ def perform_genetic_fit(
                 "The optimizer determines the number of generations, epochs will be ignored."
             )
 
-        metricas = self.metrics_names
         for epoch in range(epochs):
             # Generate the best mutant
             score, best_mutant = self.opt_instance.run_step(x=x, y=y)
 
+            training_metric = next(iter(score))
+
             # Ensure the best mutant is the current one
             self.set_weights(best_mutant)
             if verbose == 1:
-                loss = parse_eval(score)
+                loss = score[training_metric]
                 information = f" > epoch: {epoch+1}/{epochs}, {loss} "
                 log.info(information)
+
             # Fill keras history
-            try:
-                history_data = dict(zip(metricas, score))
-            except TypeError as e:
-                # Maybe score was just one number, evil Keras
-                if parse_eval(score) == score:
-                    score = [score, score]
-                    history_data = dict(zip(metricas, score))
-                else:
-                    raise e
-            self.history.on_epoch_end(epoch, history_data)
-        return self.history
+            history_data = score
+            self.history_info.on_epoch_end(epoch, history_data)
+
+        return self.history_info
 
     def fit(self, x=None, y=None, validation_data=None, epochs=1, verbose=0, **kwargs):
         """ If the optimizer is genetic, the fitting procedure consists on executing `run_step` for