Support python 3.12, Deprecate python 3.8

.github/workflows/docs.yml

@@ -14,7 +14,7 @@ jobs:
     - name: Python
       uses: actions/setup-python@v2
       with:
-        python-version: 3.8
+        python-version: '3.10'
 
     - name: Build
       run: |

.github/workflows/latest-dependencies.yml

@@ -8,10 +8,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
-    - name: Set up Python 3.8
+    - name: Set up Python 3.10
       uses: actions/setup-python@v2
       with:
-        python-version: 3.8
+        python-version: '3.10'
     - name: Install package
       run: |
         pip install --upgrade pip

.github/workflows/tests.yml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: [3.8]
+        python-version: ['3.10']
         os: [ubuntu-latest, macos-latest, windows-latest]
     steps:
     - uses: actions/checkout@v1
@@ -32,7 +32,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ['3.8', '3.9', '3.10', '3.11']
+        python-version: ['3.9', '3.10', '3.11', '3.12']
         os: [ubuntu-latest]
     steps:
     - uses: actions/checkout@v1
@@ -50,7 +50,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ['3.8', '3.9', '3.10', '3.11']
+        python-version: ['3.9', '3.10', '3.11', '3.12']
         os: [ubuntu-latest, macos-latest]
     steps:
     - uses: actions/checkout@v1
@@ -68,7 +68,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ['3.8', '3.9', '3.10', '3.11']
+        python-version: ['3.9', '3.10', '3.11', '3.12']
         os: [ubuntu-latest, macos-latest, windows-latest]
     steps:
     - uses: actions/checkout@v1
@@ -105,8 +105,8 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ['3.8']
-        os: [macos-13]
+        python-version: ['3.9']
+        os: [ubuntu-latest, macos-latest]
     steps:
     - uses: actions/checkout@v4
     - name: Set up Python ${{ matrix.python-version }}
@@ -115,7 +115,7 @@ jobs:
         python-version: ${{ matrix.python-version }}
     - name: Install package and dependencies
       run: |
-          python -m pip install --upgrade pip
+          python -m pip install --upgrade 'pip<=24.1'
           python -m pip install invoke .[test]
     - name: invoke minimum
       run: invoke minimum
@@ -125,7 +125,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ['3.8', '3.9', '3.10', '3.11']
+        python-version: ['3.9', '3.10', '3.11', '3.12']
         os: [ubuntu-latest, macos-latest, windows-latest]
         exclude:
           - os: windows-latest

README.md

@@ -8,7 +8,7 @@
 </p>
 
 [![Development Status](https://img.shields.io/badge/Development%20Status-2%20--%20Pre--Alpha-yellow)](https://pypi.org/search/?c=Development+Status+%3A%3A+2+-+Pre-Alpha)
-[![Python](https://img.shields.io/badge/Python-3.8%20%7C%203.9%20%7C%203.10%20%7C%203.11-blue)](https://badge.fury.io/py/orion-ml) 
+[![Python](https://img.shields.io/badge/Python-3.9%20%7C%203.10%20%7C%203.11%20%7C%203.12-blue)](https://badge.fury.io/py/orion-ml) 
 [![PyPi Shield](https://img.shields.io/pypi/v/orion-ml.svg)](https://pypi.python.org/pypi/orion-ml)
 [![Tests](https://github.com/sintel-dev/Orion/workflows/Run%20Tests/badge.svg)](https://github.com/sintel-dev/Orion/actions?query=workflow%3A%22Run+Tests%22+branch%3Amaster)
 [![Downloads](https://pepy.tech/badge/orion-ml)](https://pepy.tech/project/orion-ml)

docs/conf.py

@@ -31,7 +31,6 @@
 # Add any Sphinx extension module names here, as strings. They can be
 # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom ones.
 extensions = [
-    'm2r2',
     'nbsphinx',
     'sphinx.ext.autodoc',
     'sphinx.ext.autosummary',
@@ -96,7 +95,7 @@
 #
 # This is also used if you do content translation via gettext catalogs.
 # Usually you set "language" from the command line for these cases.
-language = None
+language = 'en'
 
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.

docs/user_guides/primitives_pipelines/primitives/DenseSeq2Seq.rst

@@ -29,7 +29,7 @@ argument                type                description
  ``input_shape``        ``tuple``           tuple denoting the shape of an input sample
  ``target_shape``       ``tuple``           tuple denoting the shape of an output sample
  ``optimizer``          ``str``             string (name of optimizer) or optimizer instance. Default is ``keras.optimizers.Adam``
- ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``keras.losses.mean_squared_error``
+ ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``tensorflow.losses.mse``
  ``metrics``            ``list``            list of metrics to be evaluated by the model during training and testing. Default is ["mse"]
  ``return_seqeunces``   ``bool``            whether to return the last output in the output sequence, or the full sequence. Default is False
  ``layers``             ``list``            list of keras layers which are the basic building blocks of a neural network

docs/user_guides/primitives_pipelines/primitives/LSTMSeq2Seq.rst

@@ -29,7 +29,7 @@ argument                type                description
  ``input_shape``        ``tuple``           tuple denoting the shape of an input sample
  ``target_shape``       ``tuple``           tuple denoting the shape of an output sample
  ``optimizer``          ``str``             string (name of optimizer) or optimizer instance. Default is ``keras.optimizers.Adam``
- ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``keras.losses.mean_squared_error``
+ ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``tensorflow.losses.mse``
  ``metrics``            ``list``            list of metrics to be evaluated by the model during training and testing. Default is ["mse"]
  ``return_seqeunces``   ``bool``            whether to return the last output in the output sequence, or the full sequence. Default is False
  ``layers``             ``list``            list of keras layers which are the basic building blocks of a neural network

docs/user_guides/primitives_pipelines/primitives/LSTMTimeSeriesRegressor.rst

@@ -28,7 +28,7 @@ argument                type                description
  ``input_shape``        ``tuple``           tuple denoting the shape of an input sample
  ``dense_units``        ``int``             number of values ahead to predict (target size). Default is 1.
  ``optimizer``          ``str``             string (name of optimizer) or optimizer instance. Default is ``keras.optimizers.Adam``
- ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``keras.losses.mean_squared_error``
+ ``loss``               ``str``             string (name of the objective function) or an objective function instance. Default is ``tensorflow.losses.mse``
  ``metrics``            ``list``            list of metrics to be evaluated by the model during training and testing. Default is ["mse"]
  ``return_seqeunces``   ``bool``            whether to return the last output in the output sequence, or the full sequence. Default is False
  ``layers``             ``list``            list of keras layers which are the basic building blocks of a neural network
@@ -53,7 +53,7 @@ argument                type                description
     X = np.array([1] * 100).reshape(1, -1, 1)
     y = np.array([[1]])
     primitive = load_primitive('keras.Sequential.LSTMTimeSeriesRegressor', 
-        arguments={"X": X, "y": y, "input_shape":(100, 1), "batch_size": 1, "validation_split": 0})
+        arguments={"X": X, "y": y, "input_shape":(100, 1), "batch_size": 1, "validation_split": 0, "loss": "tensorflow.losses.mse"})
 
     primitive.fit()
     primitive.produce(X=X)

orion/__main__.py

@@ -5,8 +5,6 @@
 import logging
 import warnings
 
-import tabulate
-
 from orion.analysis import get_available_templates
 from orion.benchmark import benchmark
 from orion.evaluation import CONTEXTUAL_METRICS as METRICS
@@ -27,13 +25,6 @@ def _evaluate(args):
         print('Writing results in {}'.format(args.output))
         scores.to_csv(args.output, index=False)
 
-    print(tabulate.tabulate(
-        scores,
-        showindex=False,
-        tablefmt='github',
-        headers=scores.columns
-    ))
-
 
 def get_parser():
 

orion/pipelines/verified/lstm_dynamic_threshold/lstm_dynamic_threshold.json

@@ -25,7 +25,8 @@
             "window_size": 250
         },
         "keras.Sequential.LSTMTimeSeriesRegressor": {
-            "epochs": 35
+            "epochs": 35,
+            "loss": "tensorflow.losses.mse"
         },
         "orion.primitives.timeseries_anomalies.find_anomalies#1": {
             "window_size_portion": 0.33,

orion/primitives/adapters/ncps.py

@@ -8,7 +8,7 @@
 def build_layer(layer, hyperparameters):
     layer_class = import_object(layer['class'])
     layer_kwargs = layer['parameters'].copy()
-    if issubclass(layer_class, tf.keras.layers.Wrapper):
+    if issubclass(layer_class, tf.keras.layers.Layer) and 'layer' in layer_kwargs:
         layer_kwargs['layer'] = build_layer(layer_kwargs['layer'], hyperparameters)
     elif issubclass(layer_class, tf.keras.layers.RNN) and isinstance(layer_kwargs['units'], dict):
         layer_kwargs['units'] = build_layer(layer_kwargs['units'], hyperparameters)

orion/primitives/aer.py

@@ -11,22 +11,11 @@
 from tensorflow.keras.models import Model
 
 from orion.primitives.timeseries_errors import reconstruction_errors, regression_errors
+from orion.primitives.utils import build_layer
 
 LOGGER = logging.getLogger(__name__)
 
 
-def build_layer(layer: dict, hyperparameters: dict):
-    layer_class = import_object(layer['class'])
-    layer_kwargs = layer['parameters'].copy()
-    # TODO: Upgrade to using tf.keras.layers.Wrapper in mlprimitives.
-    if issubclass(layer_class, tf.keras.layers.Wrapper):
-        layer_kwargs['layer'] = build_layer(layer_kwargs['layer'], hyperparameters)
-    for key, value in layer_kwargs.items():
-        if isinstance(value, str):
-            layer_kwargs[key] = hyperparameters.get(value, value)
-    return layer_class(**layer_kwargs)
-
-
 class AER(object):
     """Autoencoder with bi-directional regression for time series anomaly detection.
 

orion/primitives/jsons/keras.Sequential.DenseSeq2Seq.json

@@ -92,7 +92,7 @@
             },
             "loss": {
                 "type": "str",
-                "default": "tensorflow.keras.losses.mean_squared_error"
+                "default": "tensorflow.losses.mse"
             },
             "metrics": {
                 "type": "list",

orion/primitives/jsons/keras.Sequential.LSTMSeq2Seq.json

@@ -92,7 +92,7 @@
             },
             "loss": {
                 "type": "str",
-                "default": "tensorflow.keras.losses.mean_squared_error"
+                "default": "tensorflow.losses.mse"
             },
             "metrics": {
                 "type": "list",

orion/primitives/jsons/keras.Sequential.LTCTimeSeriesRegressor.json

@@ -92,7 +92,7 @@
             },
             "loss": {
                 "type": "str",
-                "default": "tensorflow.keras.losses.mean_squared_error"
+                "default": "tensorflow.losses.mse"
             },
             "metrics": {
                 "type": "list",
@@ -149,7 +149,7 @@
                         }
                     },
                     {
-                        "class": "ncps.tf.LTC",
+                        "class": "ncps.keras.LTC",
                         "parameters": {
                             "units": {
                                 "class": "ncps.wirings.AutoNCP",

orion/primitives/tadgan.py

@@ -14,6 +14,7 @@
 from tensorflow.keras import Model
 
 from orion.primitives.timeseries_errors import reconstruction_errors
+from orion.primitives.utils import build_layer
 
 LOGGER = logging.getLogger(__name__)
 tf.keras.backend.set_floatx('float64')
@@ -25,20 +26,6 @@
 ]
 
 
-def build_layer(layer: dict, hyperparameters: dict):
-    layer_class = import_object(layer['class'])
-    layer_kwargs = layer['parameters'].copy()
-    # TODO: Upgrade to using tf.keras.layers.Wrapper in mlprimitives.
-    if issubclass(layer_class, tf.keras.layers.Wrapper):
-        layer_kwargs['layer'] = build_layer(layer_kwargs['layer'], hyperparameters)
-
-    for key, value in layer_kwargs.items():
-        if isinstance(value, str):
-            layer_kwargs[key] = hyperparameters.get(value, value)
-
-    return layer_class(**layer_kwargs)
-
-
 class TadGAN:
     """TadGAN model for time series reconstruction.
 

orion/primitives/timeseries_anomalies.py

@@ -193,7 +193,7 @@ def _find_sequences(errors, epsilon, anomaly_padding):
     for idx in index_above.flatten():
         above[max(0, idx - anomaly_padding):min(idx + anomaly_padding + 1, len(above))] = True
 
-    shift = above.shift(1).fillna(False)
+    shift = above.shift(1).astype('boolean').fillna(False)
     change = above != shift
 
     if above.all():

orion/primitives/timeseries_errors.py

@@ -91,9 +91,9 @@ def _area_error(y, y_hat, score_window=10):
             An array of area error.
     """
     smooth_y = pd.Series(y).rolling(
-        score_window, center=True, min_periods=score_window // 2).apply(integrate.trapz)
+        score_window, center=True, min_periods=score_window // 2).apply(integrate.trapezoid)
     smooth_y_hat = pd.Series(y_hat).rolling(
-        score_window, center=True, min_periods=score_window // 2).apply(integrate.trapz)
+        score_window, center=True, min_periods=score_window // 2).apply(integrate.trapezoid)
 
     errors = abs(smooth_y - smooth_y_hat)
 

orion/primitives/utils.py

@@ -0,0 +1,16 @@
+# -*- coding: utf-8 -*-
+
+import tensorflow as tf
+from mlstars.utils import import_object
+
+
+def build_layer(layer: dict, hyperparameters: dict):
+    layer_class = import_object(layer['class'])
+    layer_kwargs = layer['parameters'].copy()
+    # TODO: tf.keras.layers.Wrapper deprecated, Bidirectional inheret from Layer
+    if issubclass(layer_class, tf.keras.layers.Layer) and 'layer' in layer_kwargs:
+        layer_kwargs['layer'] = build_layer(layer_kwargs['layer'], hyperparameters)
+    for key, value in layer_kwargs.items():
+        if isinstance(value, str):
+            layer_kwargs[key] = hyperparameters.get(value, value)
+    return layer_class(**layer_kwargs)

orion/primitives/vae.py

@@ -16,19 +16,17 @@
 from tensorflow.keras.layers import Input
 from tensorflow.keras.models import Model
 
+from orion.primitives.utils import build_layer
+
 LOGGER = logging.getLogger(__name__)
 
 
-def build_layer(layer: dict, hyperparameters: dict):
-    layer_class = import_object(layer['class'])
-    layer_kwargs = layer['parameters'].copy()
-    # TODO: Upgrade to using tf.keras.layers.Wrapper in mlprimitives.
-    if issubclass(layer_class, tf.keras.layers.Wrapper):
-        layer_kwargs['layer'] = build_layer(layer_kwargs['layer'], hyperparameters)
-    for key, value in layer_kwargs.items():
-        if isinstance(value, str):
-            layer_kwargs[key] = hyperparameters.get(value, value)
-    return layer_class(**layer_kwargs)
+class KLDivergenceLoss(tf.keras.layers.Layer):
+    def call(self, inputs):
+        z_log_sigma, z_mean = inputs
+        kl_loss = -0.5 * K.mean(1 + z_log_sigma - K.square(z_mean) - K.exp(z_log_sigma), axis=-1)
+        self.add_loss(kl_loss)
+        return inputs
 
 
 class VAE(object):
@@ -180,13 +178,13 @@ def _build_vae(self, **kwargs):
         h = self.encoder(x)
         z_mean = tf.keras.layers.Dense(self.latent_dim)(h)
         z_log_sigma = tf.keras.layers.Dense(self.latent_dim)(h)
+        KLDivergenceLoss()([z_log_sigma, z_mean])  # kl loss
         z = tf.keras.layers.Lambda(self._sampling)([z_mean, z_log_sigma])
 
         y_ = self.generator(z)
 
         self.vae_model = Model([x, y], y_)
-        self.vae_model.add_loss(self._vae_loss(y, y_, z_log_sigma, z_mean))
-        self.vae_model.compile(optimizer=self.optimizer)
+        self.vae_model.compile(loss='mse', optimizer=self.optimizer)
 
     def fit(self, X: np.ndarray, y: np.ndarray, **kwargs):
         """Fit the model.
@@ -211,7 +209,7 @@ def fit(self, X: np.ndarray, y: np.ndarray, **kwargs):
             for callback in self.callbacks
         ]
 
-        self.fit_history = self.vae_model.fit((X, y),
+        self.fit_history = self.vae_model.fit((X, y), y,
                                               batch_size=self.batch_size,
                                               epochs=self.epochs,
                                               shuffle=self.shuffle,