Multimodal models

docs/requirements.txt

@@ -1,13 +1,27 @@
-sphinx
-sphinx-rtd-theme
-sphinx-autodoc-typehints
+shap==0.46.0
+scikit-learn>=1.0.1
 pandas
-numpy
 torch
-matplotlib
-plotly
+tb-nightly
+seaborn
+future
+h5py
+wandb==0.17.2
+google-cloud
 google-cloud-storage
-scikit-learn
-wandb
-shap
+plotly~=5.24.0
+pytz>=2022.1
+setuptools~=69.5.1
+numpy==1.26
+requests
+torchvision>=0.6.0
+mpld3>=0.5
+numba>=0.50
+sphinx
+sphinx-rtd-theme
+sphinx-autodoc-typehints
+sphinx
 einops
+pytorch-tsmixer
+einsum
+jaxtyping

docs/source/basic_ae.rst

@@ -1,5 +1,5 @@
-Simple AE
-==============
+Simple Auto Encoder
+==================
 
 .. automodule:: flood_forecast.meta_models.basic_ae
     :members:

docs/source/basic_utils.rst

@@ -1,5 +1,5 @@
 Basic Google Cloud Platform Utilities
-=====================================
+======================================
 
 Flow Forecast natively integrates with Google Cloud Platform.
 

docs/source/custom_opt.rst

@@ -1,5 +1,5 @@
-Custom Optimizers and more
-==========================
+Custom Optimizers
+====================
 
 .. automodule:: flood_forecast.custom.custom_opt
     :members:

flood_forecast/basic/base_line_methods.py

@@ -18,13 +18,13 @@ def __init__(self, seq_length: int, n_time_series: int, output_seq_len=1, metric
         self.metric_function = self.metric_dict[metric]
 
     def forward(self, x: torch.Tensor) -> torch.Tensor:
-        """_summary_
+        """The forward pass of the NaiveModel
 
-        Args:
-            x (torch.Tensor): _description_
+        param x: The input tensor of shape (batch_size, seq_len, n_time_series)
+        type x: torch.Tensor
 
         Returns:
-            torch.Tensor: _description_
+            torch.Tensor: The output tensor of shape (batch_size, output_seq_len, 1)
         """
         return self.metric_function(x, self.output_seq_len)
 

flood_forecast/basic/d_n_linear.py

@@ -3,7 +3,7 @@
 
 
 class NLinear(nn.Module):
-    """Normalization-Linear."""
+    """Normalization of Linear fixing."""
 
     def __init__(self, forecast_history: int, forecast_length: int, enc_in=128, individual=False, n_targs=1):
         super(NLinear, self).__init__()
@@ -113,11 +113,11 @@ def __init__(self, forecast_history: int, forecast_length: int, individual, enc_
             # self.Linear_Trend.weight = nn.Parameter((1/self.seq_len)*torch.ones([self.pred_len2,self.seq_len]))
 
     def forward(self, x: torch.Tensor):
-        """The.
+        """The forward pass of the D-Linear model.
 
         :param x: PyTorch tensor of size [Batch, Input length, Channel]
         :type x: _type_
-        :return: _description_
+        :return: Returns the output tensor of size [Batch, Output length, Channel]
         :rtype: _type_
         """
         seasonal_init, trend_init = self.decompsition(x)

flood_forecast/basic/gru_vanilla.py

@@ -8,10 +8,10 @@ def __init__(self, n_time_series: int, hidden_dim: int, num_layers: int, n_targe
 
         :param n_time_series: The number of time series present in the data
         :type n_time_series int:
-        :param hidden_dim:
+        :param hidden_dim: The hidden dimension of the GRU.
         :type hidden_dim:
 
-        Note for probablistic n_targets must be set to two and actual multiple targs are not supported now.
+        Note for probabilistic n_targets must be set to two and actual multiple targets are not supported now.
         """
         super(VanillaGRU, self).__init__()
 

flood_forecast/evaluator.py

@@ -31,7 +31,8 @@
 def stream_baseline(
     river_flow_df: pd.DataFrame, forecast_column: str, hours_forecast=336
 ) -> Tuple[pd.DataFrame, float]:
-    """Function to compute the baseline MSE by using the mean value from the train data."""
+    """Function to compute the baseline MSE by using the mean value from the train data.
+    """
     total_length = len(river_flow_df.index)
     train_river_data = river_flow_df[: total_length - hours_forecast]
     test_river_data = river_flow_df[total_length - hours_forecast:]
@@ -162,8 +163,9 @@ def evaluate_model(
     return eval_log, df_train_and_test, forecast_start_idx, df_predictions
 
 
-def run_evaluation(model, df_train_and_test, forecast_history, target_col, end_tensor, g_loss=False, eval_log={},
-                   end_tensor_0=None) -> Dict:
+def run_evaluation(model, df_train_and_test: pd.DataFrame, forecast_history, target_col: str, end_tensor, g_loss=False,
+                   eval_log: Dict = {}, end_tensor_0=None) -> Dict:
+    """Function to run evaluation on the model."""
     inference_params = model.params["inference_params"]
     for evaluation_metric in model.crit:
         idx = 0

flood_forecast/explain_model_output.py

@@ -18,7 +18,7 @@
 
 def handle_dl_output(dl, dl_class: str, datetime_start: datetime, device: str) -> Tuple[torch.Tensor, int]:
     """
-    :param dl: The test data-loader. Should be passed directly
+    :param dl: The test data-loader. Should be passed directly.
     :type dl: Union[CSVTestLoader, TemporalTestLoader]
     :param dl_class: A string that is the name of DL passef from the params file
     :type dl_class: str