fix docs

docs/index.rst

@@ -1,6 +1,6 @@
 :github_url: https://github.com/
 
-Time Series Simulator (TSGM) Official Documentation
+Time Series Generative Modeling (TSGM) Official Documentation
 ========================================
 
 Time Series Generative Modeling (TSGM) is a Python framework for time series data generation. It include data-driven and model-based approaches to synthetic time-series generation. It uses both generative

tsgm/models/augmentations.py

@@ -412,7 +412,7 @@ def generate(
             Initial timesteries to start from for the optimization process, with shape (original_size, d).
             In case y is given, the shape of initial_timeseries is assumed to be (n_classes, original_size, d)
         initial_labels: array or None (default: None)
-            TODO
+            Labels for samples from `initial_timeseries`
         Returns
         -------
         np.array of shape (n_samples, original_size, d) if y is None
@@ -437,8 +437,8 @@ def generate(
 
             y_new = []
             X_new = []
-            unique_lables = np.unique(initial_labels)
-            for i, label in enumerate(unique_lables):
+            unique_labels = np.unique(initial_labels)
+            for i, label in enumerate(unique_labels):
                 logger.debug(f"DTWBA Class {label}...")
                 cur_initial_timeseries = initial_timeseries[np.ravel(initial_labels) == label]
                 n_samples_per_label = len(cur_initial_timeseries)

tsgm/models/cgan.py

@@ -85,8 +85,12 @@ def _get_random_vector_labels(self, batch_size: int, labels=None) -> tsgm.types.
     def train_step(self, data: tsgm.types.Tensor) -> T.Dict[str, float]:
         """
         Performs a training step using a batch of data, stored in data.
+
         :param data: A batch of data in a format batch_size x seq_len x feat_dim
         :type data: tsgm.types.Tensor
+
+        :returns: A dictionary with generator (key "g_loss") and discriminator (key "d_loss") losses
+        :rtype: T.Dict[str, float]
         """
         real_data = data
         batch_size = tf.shape(real_data)[0]
@@ -140,11 +144,20 @@ def generate(self, num: int) -> tsgm.types.Tensor:
 
         :param num: the number of samples to be generated.
         :type num: int
+
+        :returns: Generated samples
+        :rtype: tsgm.types.Tensor
         """
         random_vector_labels = self._get_random_vector_labels(batch_size=num)
         return self.generator(random_vector_labels)
 
     def clone(self) -> "GAN":
+        """
+        Clones GAN object
+
+        :returns: The exact copy of the object
+        :rtype: "GAN"
+        """
         copy_model = GAN(self.discriminator, self.generator, latent_dim=self.latent_dim)
         copy_model = copy_model.set_weights(self.get_weights())
         return copy_model
@@ -181,6 +194,7 @@ def __init__(self, discriminator: keras.Model, generator: keras.Model, latent_di
     def metrics(self) -> T.List:
         """
         :returns: A list of metrics trackers (e.g., generator's loss and discriminator's loss).
+        :rtype: T.List
         """
         return [self.gen_loss_tracker, self.disc_loss_tracker]
 
@@ -233,14 +247,13 @@ def _get_output_shape(self, labels: tsgm.types.Tensor) -> int:
 
     def train_step(self, data: T.Tuple) -> T.Dict[str, float]:
         """
-        Compiles the generator and discriminator models.
+        Performs a training step using a batch of data, stored in data.
 
-        :param d_optimizer: An optimizer for the GAN's discriminator.
-        :type d_optimizer: keras.Model
-        :param g_optimizer: An optimizer for the GAN's generator.
-        :type generator: keras.Model
-        :param loss_fn: Loss function.
-        :type loss_fn: keras.losses.Loss
+        :param data: A batch of data in a format batch_size x seq_len x feat_dim
+        :type data: tsgm.types.Tensor
+
+        :returns: A dictionary with generator (key "g_loss") and discriminator (key "d_loss") losses
+        :rtype: T.Dict[str, float]
         """
         real_ts, labels = data
         output_dim = self._get_output_shape(labels)
@@ -319,6 +332,9 @@ def generate(self, labels: tsgm.types.Tensor) -> tsgm.types.Tensor:
 
         :param labels: the number of samples to be generated.
         :type labels: tsgm.types.Tensor
+
+        :returns: generated samples
+        :rtype: tsgm.types.Tensor
         """
         batch_size = labels.shape[0]
 

tsgm/models/cvae.py

@@ -46,8 +46,12 @@ def metrics(self) -> T.List:
     def call(self, X: tsgm.types.Tensor) -> tsgm.types.Tensor:
         """
         Encodes and decodes time series dataset X.
+
         :param X: The size of the noise vector.
         :type X: tsgm.types.Tensor
+
+        :returns: Generated samples
+        :rtype: tsgm.types.Tensor
         """
         z_mean, _, _ = self.encoder(X)
         x_decoded = self.decoder(z_mean)
@@ -64,10 +68,12 @@ def _get_reconstruction_loss(self, X: tsgm.types.Tensor, Xr: tsgm.types.Tensor)
     def train_step(self, data: tsgm.types.Tensor) -> T.Dict:
         """
         Performs a training step using a batch of data, stored in data.
+
         :param data: A batch of data in a format batch_size x seq_len x feat_dim
         :type data: tsgm.types.Tensor
 
-        :returns a dict with losses:
+        :returns: A dict with losses
+        :rtype: T.Dict
         """
         with tf.GradientTape() as tape:
             z_mean, z_log_var, z = self.encoder(data)
@@ -94,7 +100,8 @@ def generate(self, n: int) -> tsgm.types.Tensor:
         :param n: the number of samples to be generated.
         :type n: int
 
-        :returns: a tensor with generated samples.
+        :returns: A tensor with generated samples.
+        :rtype: tsgm.types.Tensor
         """
         z = tf.random.normal((n, self.latent_dim))
         return self.decoder(z)
@@ -135,6 +142,7 @@ def generate(self, labels: tsgm.types.Tensor) -> T.Tuple[tsgm.types.Tensor, tsgm
         :type labels: tsgm.types.Tensor
 
         :returns: a tuple of synthetically generated data and labels.
+        :rtype: T.Tuple[tsgm.types.Tensor, tsgm.types.Tensor]
         """
         batch_size = tf.shape(labels)[0]
         z = tf.random.normal((batch_size, self._seq_len, self.latent_dim), dtype=labels.dtype)
@@ -144,8 +152,12 @@ def generate(self, labels: tsgm.types.Tensor) -> T.Tuple[tsgm.types.Tensor, tsgm
     def call(self, data: tsgm.types.Tensor) -> tsgm.types.Tensor:
         """
         Encodes and decodes time series dataset X.
+
         :param X: The size of the noise vector.
         :type X: tsgm.types.Tensor
+
+        :returns: Generated samples
+        :rtype: tsgm.types.Tensor
         """
         X, labels = data
         encoder_input = self._get_encoder_input(X, labels)
@@ -180,10 +192,12 @@ def _get_decoder_input(self, z: tsgm.types.Tensor, labels: tsgm.types.Tensor) ->
     def train_step(self, data: tsgm.types.Tensor) -> T.Dict[str, float]:
         """
         Performs a training step using a batch of data, stored in data.
+
         :param data: A batch of data in a format batch_size x seq_len x feat_dim
         :type data: tsgm.types.Tensor
 
-        :returns a dict with losses:
+        :returns: A dict with losses
+        :rtype: T.Dict[str, float]
         """
         X, labels = data
         with tf.GradientTape() as tape:

tsgm/optimization/abc.py

@@ -56,9 +56,10 @@ def _calc_statistics(self, data: tsgm.dataset.Dataset) -> tsgm.types.Tensor:
     def sample_parameters(self, n_samples: int) -> T.List:
         """
         Samples parameters from the rejection sampler.
+
         :param n_samples: Number of samples
         :type simulator: int
-        :return: A list of samples. Each sample is represent as dict.
+        :returns: A list of samples. Each sample is represent as dict.
         :rtype: T.List[T.Dict]
         """
         cur_sim = self._simulator.clone()
@@ -78,7 +79,7 @@ def sample_parameters(self, n_samples: int) -> T.List:
         return samples
 
 
-def prior_samples(priors: T.Dict, params: T.List) -> T.List:
+def prior_samples(priors: T.Dict, params: T.List) -> T.Dict:
     """
     Generate prior samples for the specified parameters.
 

tsgm/version.py

@@ -1 +1 @@
-__version__ = "0.0.4"
+__version__ = "0.0.4"