diff --git a/pipelines/precipitation_model/impa/flows.py b/pipelines/precipitation_model/impa/flows.py
index e422c7b..42853b6 100644
--- a/pipelines/precipitation_model/impa/flows.py
+++ b/pipelines/precipitation_model/impa/flows.py
@@ -66,15 +66,17 @@
         "start_datetime",
         default=None,
         required=False,
-        #description="Datetime in YYYY-MM-dd HH:mm:ss format, UTC timezone",
+        # description="Datetime in YYYY-MM-dd HH:mm:ss format, UTC timezone",
     )
     num_workers = Parameter(
         "num_workers",
         default=8,
         required=False,
-        #description="Number of workers to use for parallel processing",
+        # description="Number of workers to use for parallel processing",
     )
-    cuda = Parameter("cuda", default=False, required=False)  #, description="Use CUDA for prediction"
+    cuda = Parameter(
+        "cuda", default=False, required=False
+    )  # , description="Use CUDA for prediction"
 
     # Parameters for saving data on GCP
     materialize_after_dump = Parameter("materialize_after_dump", default=False, required=False)
diff --git a/pipelines/precipitation_model/impa/src/data/HDFDataset2.py b/pipelines/precipitation_model/impa/src/data/HDFDataset2.py
index 2a3dcb0..7f7b38e 100644
--- a/pipelines/precipitation_model/impa/src/data/HDFDataset2.py
+++ b/pipelines/precipitation_model/impa/src/data/HDFDataset2.py
@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 from pathlib import Path
 
 import h5py
@@ -5,8 +6,16 @@
 import torch
 from torch.utils import data
 
-from pipelines.precipitation_model.impa.src.utils.dataframe_utils import N_AFTER, N_BEFORE, fetch_future_datetimes, fetch_reversed_past_datetimes
-from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok, print_warning
+from pipelines.precipitation_model.impa.src.utils.dataframe_utils import (
+    N_AFTER,
+    N_BEFORE,
+    fetch_future_datetimes,
+    fetch_reversed_past_datetimes,
+)
+from pipelines.precipitation_model.impa.src.utils.general_utils import (
+    print_ok,
+    print_warning,
+)
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import get_dataset_keys
 
 MIN_WEIGHT = 100
@@ -56,16 +65,27 @@ def __init__(
         else:
             suffix = ""
 
-        if len(set(["latent_field", "motion_field", "intensities"]).intersection(set(get_item_output))) > 0:
+        if (
+            len(
+                set(["latent_field", "motion_field", "intensities"]).intersection(
+                    set(get_item_output)
+                )
+            )
+            > 0
+        ):
             if autoencoder_hash is None:
                 autoencoder_hash = "001178e117f50cf17817f336b86a809f"
             parent_path = Path(filepath).parents[0]
             if "train" in filepath.stem:
-                self.latent_field_filepath = Path(f"{parent_path}/train_latent{suffix}_{autoencoder_hash}.hdf")
+                self.latent_field_filepath = Path(
+                    f"{parent_path}/train_latent{suffix}_{autoencoder_hash}.hdf"
+                )
                 self.motion_field_filepath = Path(f"{parent_path}/train_motion{suffix}.hdf")
                 self.intensities_filepath = Path(f"{parent_path}/train_intensities{suffix}.hdf")
             elif "val" in filepath.stem:
-                self.latent_field_filepath = Path(f"{parent_path}/val_latent{suffix}_{autoencoder_hash}.hdf")
+                self.latent_field_filepath = Path(
+                    f"{parent_path}/val_latent{suffix}_{autoencoder_hash}.hdf"
+                )
                 self.motion_field_filepath = Path(f"{parent_path}/val_motion{suffix}.hdf")
                 self.intensities_filepath = Path(f"{parent_path}/val_intensities{suffix}.hdf")
             else:
@@ -267,9 +287,12 @@ def __len__(self):
             return len(self.keys)
 
     def get_sample_weights(self, overwrite_if_exists=False, verbose=True):
-        assert self.filepath.stem == "train", "Sample weights can only be calculated for the train dataset."
+        assert (
+            self.filepath.stem == "train"
+        ), "Sample weights can only be calculated for the train dataset."
         weights_filepath = (
-            self.filepath.parents[0] / f"train_sample_weights2-n_before={self.n_before}-n_after={self.n_after}.npy"
+            self.filepath.parents[0]
+            / f"train_sample_weights2-n_before={self.n_before}-n_after={self.n_after}.npy"
         )
         if not overwrite_if_exists:
             if weights_filepath.is_file():
diff --git a/pipelines/precipitation_model/impa/src/data/HDFDatasetLocations.py b/pipelines/precipitation_model/impa/src/data/HDFDatasetLocations.py
index b106717..0f08a7b 100644
--- a/pipelines/precipitation_model/impa/src/data/HDFDatasetLocations.py
+++ b/pipelines/precipitation_model/impa/src/data/HDFDatasetLocations.py
@@ -1,7 +1,13 @@
+# -*- coding: utf-8 -*-
 import numpy as np
 
-from pipelines.precipitation_model.impa.src.data.HDFDatasetMultiple import HDFDatasetMultiple
-from pipelines.precipitation_model.impa.src.utils.dataframe_utils import N_AFTER, N_BEFORE
+from pipelines.precipitation_model.impa.src.data.HDFDatasetMultiple import (
+    HDFDatasetMultiple,
+)
+from pipelines.precipitation_model.impa.src.utils.dataframe_utils import (
+    N_AFTER,
+    N_BEFORE,
+)
 
 elevation_file_small = "data/processed/elevations_data/elevation_{location}-res=2km-256x256.npy"
 elevation_file_large = "data/processed/elevations_data/elevation_{location}-res=4km-256x256.npy"
diff --git a/pipelines/precipitation_model/impa/src/data/HDFDatasetMerged.py b/pipelines/precipitation_model/impa/src/data/HDFDatasetMerged.py
index b8ecf51..abe6d34 100644
--- a/pipelines/precipitation_model/impa/src/data/HDFDatasetMerged.py
+++ b/pipelines/precipitation_model/impa/src/data/HDFDatasetMerged.py
@@ -1,7 +1,13 @@
+# -*- coding: utf-8 -*-
 import numpy as np
 
-from pipelines.precipitation_model.impa.src.data.HDFDatasetMultiple import HDFDatasetMultiple
-from pipelines.precipitation_model.impa.src.utils.dataframe_utils import N_AFTER, N_BEFORE
+from pipelines.precipitation_model.impa.src.data.HDFDatasetMultiple import (
+    HDFDatasetMultiple,
+)
+from pipelines.precipitation_model.impa.src.utils.dataframe_utils import (
+    N_AFTER,
+    N_BEFORE,
+)
 
 # For Rio de Janeiro only
 
diff --git a/pipelines/precipitation_model/impa/src/data/HDFDatasetMultiple.py b/pipelines/precipitation_model/impa/src/data/HDFDatasetMultiple.py
index 8cca6f4..a2b422f 100644
--- a/pipelines/precipitation_model/impa/src/data/HDFDatasetMultiple.py
+++ b/pipelines/precipitation_model/impa/src/data/HDFDatasetMultiple.py
@@ -8,7 +8,10 @@
 import torch
 from torch.utils import data
 
-from pipelines.precipitation_model.impa.src.utils.dataframe_utils import fetch_future_datetimes, fetch_reversed_past_datetimes
+from pipelines.precipitation_model.impa.src.utils.dataframe_utils import (
+    fetch_future_datetimes,
+    fetch_reversed_past_datetimes,
+)
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import get_dataset_keys
 
 
@@ -76,7 +79,9 @@ def __init__(
                 # print_warning("File not found in /dev/shm, using original path.")
                 new_dataframe_filepaths_array[i] = str(filepath)
 
-        self.dataframe_shm_filepaths_array = new_dataframe_filepaths_array.reshape(dataframe_filepaths_array.shape)
+        self.dataframe_shm_filepaths_array = new_dataframe_filepaths_array.reshape(
+            dataframe_filepaths_array.shape
+        )
         self.dataframe_filepaths_array = dataframe_filepaths_array
         self.n_before_array = n_before_array
         self.n_after_array = n_after_array
@@ -126,7 +131,9 @@ def _load_keys(self):
                         if j == 0:
                             new_keys = set(hdf["split_info"]["split_datetime_keys"])
                         else:
-                            new_keys = set(hdf["split_info"]["split_datetime_keys"]).intersection(new_keys)
+                            new_keys = set(hdf["split_info"]["split_datetime_keys"]).intersection(
+                                new_keys
+                            )
                     else:
                         if j == 0:
                             new_keys = set(get_dataset_keys(hdf))
@@ -140,17 +147,30 @@ def _load_keys(self):
                 self.past_keys = np.vstack(
                     [
                         self.past_keys,
-                        np.array(fetch_reversed_past_datetimes(new_keys, self.n_before_array.max(), timestep)),
+                        np.array(
+                            fetch_reversed_past_datetimes(
+                                new_keys, self.n_before_array.max(), timestep
+                            )
+                        ),
                     ]
                 )
             except ValueError:
-                self.past_keys = np.array(fetch_reversed_past_datetimes(new_keys, self.n_before_array.max(), timestep))
+                self.past_keys = np.array(
+                    fetch_reversed_past_datetimes(new_keys, self.n_before_array.max(), timestep)
+                )
             try:
                 self.future_keys = np.vstack(
-                    [self.future_keys, np.array(fetch_future_datetimes(new_keys, self.n_after_array.max(), timestep))]
+                    [
+                        self.future_keys,
+                        np.array(
+                            fetch_future_datetimes(new_keys, self.n_after_array.max(), timestep)
+                        ),
+                    ]
                 )
             except ValueError:
-                self.future_keys = np.array(fetch_future_datetimes(new_keys, self.n_after_array.max(), timestep))
+                self.future_keys = np.array(
+                    fetch_future_datetimes(new_keys, self.n_after_array.max(), timestep)
+                )
             try:
                 self.ds_indices.append(len(new_keys) + self.ds_indices[-1])
             except IndexError:
@@ -162,11 +182,21 @@ def _get_hdf_index(self, index):
                 return i
 
     def __getitem__(self, index):
-        X = torch.ones((self.ni, self.nj, (self.n_before_array * self.n_before_resolution_array).sum())) * torch.inf
+        X = (
+            torch.ones(
+                (self.ni, self.nj, (self.n_before_array * self.n_before_resolution_array).sum())
+            )
+            * torch.inf
+        )
         if self.leadtime_conditioning:
             pass
         else:
-            Y = torch.ones((self.ni, self.nj, (self.n_after_array * self.n_after_resolution_array).sum())) * torch.inf
+            Y = (
+                torch.ones(
+                    (self.ni, self.nj, (self.n_after_array * self.n_after_resolution_array).sum())
+                )
+                * torch.inf
+            )
 
         if self.leadtime_conditioning:
             leadtime_index = index % self.n_after_array[0]
@@ -186,17 +216,24 @@ def __getitem__(self, index):
                     )
                     try:
                         X[:, :, tensor_ind : tensor_ind + n_resolution] = torch.as_tensor(
-                            np.array(hdf[self.past_keys[index][i]]).reshape((self.ni, self.nj, n_resolution))
+                            np.array(hdf[self.past_keys[index][i]]).reshape(
+                                (self.ni, self.nj, n_resolution)
+                            )
                         )
                     except KeyError:
-                        X[:, :, tensor_ind : tensor_ind + n_resolution] = torch.ones((self.ni, self.nj, 1)) * np.nan
+                        X[:, :, tensor_ind : tensor_ind + n_resolution] = (
+                            torch.ones((self.ni, self.nj, 1)) * np.nan
+                        )
                 if self.leadtime_conditioning and j == 0:
                     try:
-                        Y = torch.as_tensor(np.array(hdf[self.future_keys[index][leadtime_index]])).reshape(
-                            (self.ni, self.nj, self.n_after_resolution_array[0])
-                        )
+                        Y = torch.as_tensor(
+                            np.array(hdf[self.future_keys[index][leadtime_index]])
+                        ).reshape((self.ni, self.nj, self.n_after_resolution_array[0]))
                     except KeyError:
-                        Y = torch.ones((self.ni, self.nj, self.n_after_resolution_array[0])) * np.nan
+                        Y = (
+                            torch.ones((self.ni, self.nj, self.n_after_resolution_array[0]))
+                            * np.nan
+                        )
                 else:
                     # for i, key in enumerate(self.future_keys[index]):
                     n_resolution = self.n_after_resolution_array[j]
@@ -204,7 +241,9 @@ def __getitem__(self, index):
                         if j == 0:
                             cumsum = 0
                         else:
-                            cumsum = np.cumsum(self.n_after_array * self.n_after_resolution_array)[j - 1]
+                            cumsum = np.cumsum(self.n_after_array * self.n_after_resolution_array)[
+                                j - 1
+                            ]
                         tensor_ind = cumsum + i * n_resolution
                         try:
                             Y[:, :, tensor_ind : tensor_ind + n_resolution] = torch.as_tensor(
@@ -212,7 +251,9 @@ def __getitem__(self, index):
                             ).reshape((self.ni, self.nj, n_resolution))
 
                         except KeyError:
-                            Y[:, :, tensor_ind : tensor_ind + n_resolution] = torch.ones((self.ni, self.nj, 1)) * np.nan
+                            Y[:, :, tensor_ind : tensor_ind + n_resolution] = (
+                                torch.ones((self.ni, self.nj, 1)) * np.nan
+                            )
         if self.x_transform:
             X = self.x_transform(X)
         if self.y_transform:
@@ -225,7 +266,9 @@ def __getitem__(self, index):
         day = int(date[6:8])
         hour = int(date[9:11])
         minute = int(date[11:13])
-        date = torch.tensor([month / 12, day / 31, hour / 24, minute / 60], dtype=torch.float32).reshape((1, 1, -1))
+        date = torch.tensor(
+            [month / 12, day / 31, hour / 24, minute / 60], dtype=torch.float32
+        ).reshape((1, 1, -1))
         date_tensor = date.expand((self.ni, self.nj, -1))
 
         metadata_tensor = torch.cat(
@@ -259,7 +302,9 @@ def get_sample_weights(self, min_sum=0):
             pre_weights_size = len(pre_weights)
             pre_weights = np.append(pre_weights, np.array([pre_weights.min()]))
             inds = np.load(inds_filepath).reshape(-1, 1)
-            deltas = np.arange(-self.n_before_array[0] + 1, self.n_after_array[0] + 1, dtype=int).reshape(1, -1)
+            deltas = np.arange(
+                -self.n_before_array[0] + 1, self.n_after_array[0] + 1, dtype=int
+            ).reshape(1, -1)
             slices = inds + deltas
             slices[np.logical_or(slices < 0, slices >= pre_weights_size)] = pre_weights_size
             summed_weights = pre_weights[slices].sum(axis=1).reshape(-1)
diff --git a/pipelines/precipitation_model/impa/src/data/PredHDFDataset2.py b/pipelines/precipitation_model/impa/src/data/PredHDFDataset2.py
index 0833bf6..b72f730 100644
--- a/pipelines/precipitation_model/impa/src/data/PredHDFDataset2.py
+++ b/pipelines/precipitation_model/impa/src/data/PredHDFDataset2.py
@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 from pathlib import Path
 
 import h5py
@@ -12,7 +13,10 @@
     fetch_pred_keys,
     fetch_reversed_past_datetimes,
 )
-from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok, print_warning
+from pipelines.precipitation_model.impa.src.utils.general_utils import (
+    print_ok,
+    print_warning,
+)
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import get_dataset_keys
 
 MIN_WEIGHT = 100
@@ -59,7 +63,9 @@ def __init__(
                 f"models/{model}/predictions/{dataset}/predict_{split}-ckpt={ckpt_file.replace('.ckpt','')}.hdf"
             )
         else:
-            self.predict_filepath = Path(f"models/{model}/predictions/{dataset}/predict_{split}.hdf")
+            self.predict_filepath = Path(
+                f"models/{model}/predictions/{dataset}/predict_{split}.hdf"
+            )
         self.n_predictions = n_predictions
         self.n_before = n_before
         self.n_after = n_after
@@ -69,16 +75,27 @@ def __init__(
 
         self._load_keys()
 
-        if len(set(["latent_field", "motion_field", "intensities"]).intersection(set(get_item_output))) > 0:
+        if (
+            len(
+                set(["latent_field", "motion_field", "intensities"]).intersection(
+                    set(get_item_output)
+                )
+            )
+            > 0
+        ):
             if autoencoder_hash is None:
                 autoencoder_hash = "001178e117f50cf17817f336b86a809f"
             parent_path = filepath.parents[0]
             if "train" in filepath.stem:
-                self.latent_field_filepath = Path(f"{parent_path}/train_latent_{autoencoder_hash}.hdf")
+                self.latent_field_filepath = Path(
+                    f"{parent_path}/train_latent_{autoencoder_hash}.hdf"
+                )
                 self.motion_field_filepath = Path(f"{parent_path}/train_motion.hdf")
                 self.intensities_filepath = Path(f"{parent_path}/train_intensities.hdf")
             elif "val" in filepath.stem:
-                self.latent_field_filepath = Path(f"{parent_path}/val_latent_{autoencoder_hash}.hdf")
+                self.latent_field_filepath = Path(
+                    f"{parent_path}/val_latent_{autoencoder_hash}.hdf"
+                )
                 self.motion_field_filepath = Path(f"{parent_path}/val_motion.hdf")
                 self.intensities_filepath = Path(f"{parent_path}/val_intensities.hdf")
             else:
@@ -287,9 +304,12 @@ def __len__(self):
         return len(self.keys)
 
     def get_sample_weights(self, overwrite_if_exists=False, verbose=True):
-        assert self.filepath.stem == "train", "Sample weights can only be calculated for the train dataset."
+        assert (
+            self.filepath.stem == "train"
+        ), "Sample weights can only be calculated for the train dataset."
         weights_filepath = (
-            self.filepath.parents[0] / f"train_sample_weights2-n_before={self.n_before}-n_after={self.n_after}.npy"
+            self.filepath.parents[0]
+            / f"train_sample_weights2-n_before={self.n_before}-n_after={self.n_after}.npy"
         )
         if not overwrite_if_exists:
             if weights_filepath.is_file():
diff --git a/pipelines/precipitation_model/impa/src/data/PredHDFDatasetLocations.py b/pipelines/precipitation_model/impa/src/data/PredHDFDatasetLocations.py
index 034776c..c962ba3 100644
--- a/pipelines/precipitation_model/impa/src/data/PredHDFDatasetLocations.py
+++ b/pipelines/precipitation_model/impa/src/data/PredHDFDatasetLocations.py
@@ -121,22 +121,39 @@ def _load_keys(self):
                 self.keys = np.append(self.keys, new_keys)
                 try:
                     self.past_keys = np.vstack(
-                        [self.past_keys, np.array(fetch_reversed_past_datetimes(new_keys, self.n_before, timestep))]
+                        [
+                            self.past_keys,
+                            np.array(
+                                fetch_reversed_past_datetimes(new_keys, self.n_before, timestep)
+                            ),
+                        ]
                     )
                 except ValueError:
-                    self.past_keys = np.array(fetch_reversed_past_datetimes(new_keys, self.n_before, timestep))
+                    self.past_keys = np.array(
+                        fetch_reversed_past_datetimes(new_keys, self.n_before, timestep)
+                    )
                 try:
                     self.future_keys = np.vstack(
-                        [self.future_keys, np.array(fetch_future_datetimes(new_keys, self.n_after, timestep))]
+                        [
+                            self.future_keys,
+                            np.array(fetch_future_datetimes(new_keys, self.n_after, timestep)),
+                        ]
                     )
                 except ValueError:
-                    self.future_keys = np.array(fetch_future_datetimes(new_keys, self.n_after, timestep))
+                    self.future_keys = np.array(
+                        fetch_future_datetimes(new_keys, self.n_after, timestep)
+                    )
                 try:
                     self.pred_keys = np.vstack(
-                        [self.pred_keys, np.array(fetch_pred_keys(new_keys, self.n_predictions, timestep))]
+                        [
+                            self.pred_keys,
+                            np.array(fetch_pred_keys(new_keys, self.n_predictions, timestep)),
+                        ]
                     )
                 except ValueError:
-                    self.pred_keys = np.array(fetch_pred_keys(new_keys, self.n_predictions, timestep))
+                    self.pred_keys = np.array(
+                        fetch_pred_keys(new_keys, self.n_predictions, timestep)
+                    )
                 try:
                     self.ds_indices.append(len(new_keys) + self.ds_indices[-1])
                 except IndexError:
@@ -163,11 +180,13 @@ def __getitem__(self, index):
             for i, key in enumerate(self.past_keys[index]):
                 tensor_ind = 2 * self.n_before - 2 * i - 2
                 try:
-                    X[:, :, tensor_ind : tensor_ind + 2] = torch.as_tensor(np.array(hdf[key])).reshape(
-                        (self.ni, self.nj, 2)
-                    )
+                    X[:, :, tensor_ind : tensor_ind + 2] = torch.as_tensor(
+                        np.array(hdf[key])
+                    ).reshape((self.ni, self.nj, 2))
                 except KeyError:
-                    X[:, :, tensor_ind : tensor_ind + 2] = torch.ones((self.ni, self.nj, 2)) * np.nan
+                    X[:, :, tensor_ind : tensor_ind + 2] = (
+                        torch.ones((self.ni, self.nj, 2)) * np.nan
+                    )
             if self.leadtime_conditioning:
                 try:
                     Y = torch.as_tensor(np.array(hdf[self.future_keys[index][leadtime_index]]))
@@ -176,7 +195,9 @@ def __getitem__(self, index):
             else:
                 for i, key in enumerate(self.future_keys[index]):
                     try:
-                        Y[:, :, 2 * i : 2 * i + 2] = torch.as_tensor(np.array(hdf[key])).reshape((self.ni, self.nj, 2))
+                        Y[:, :, 2 * i : 2 * i + 2] = torch.as_tensor(np.array(hdf[key])).reshape(
+                            (self.ni, self.nj, 2)
+                        )
                     except KeyError:
                         Y[:, :, 2 * i : 2 * i + 2] = torch.ones((self.ni, self.nj, 2)) * np.nan
         with h5py.File(self.predict_filepaths[hdf_index], "r") as pred_hdf:
@@ -198,7 +219,9 @@ def __getitem__(self, index):
         day = int(date[6:8])
         hour = int(date[9:11])
         minute = int(date[11:13])
-        date = torch.tensor([month / 12, day / 31, hour / 24, minute / 60], dtype=torch.float32).reshape((1, 1, -1))
+        date = torch.tensor(
+            [month / 12, day / 31, hour / 24, minute / 60], dtype=torch.float32
+        ).reshape((1, 1, -1))
         date_tensor = date.expand((self.ni, self.nj, -1))
 
         metadata_tensor = torch.cat(
diff --git a/pipelines/precipitation_model/impa/src/data/process/SatelliteData.py b/pipelines/precipitation_model/impa/src/data/process/SatelliteData.py
index e73a25b..ec0541e 100644
--- a/pipelines/precipitation_model/impa/src/data/process/SatelliteData.py
+++ b/pipelines/precipitation_model/impa/src/data/process/SatelliteData.py
@@ -48,7 +48,9 @@ def load_data(cls, input_filepath: Path | str, value: str | None = None):
     def _load_previous_day(self):
         self.data = pd.concat(
             [
-                pd.read_feather(f"{self.folder}/{self.product}/{self.day - timedelta(days=1)}.feather"),
+                pd.read_feather(
+                    f"{self.folder}/{self.product}/{self.day - timedelta(days=1)}.feather"
+                ),
                 self.data,
             ]
         )
@@ -57,7 +59,9 @@ def _load_cloud_height(self) -> pd.DataFrame:
         return pd.concat(
             [
                 pd.read_feather(f"{self.folder}/ABI-L2-ACHAF/{self.day}.feather"),
-                pd.read_feather(f"{self.folder}/ABI-L2-ACHAF/{self.day - timedelta(days=1)}.feather"),
+                pd.read_feather(
+                    f"{self.folder}/ABI-L2-ACHAF/{self.day - timedelta(days=1)}.feather"
+                ),
             ]
         )
 
@@ -84,7 +88,9 @@ def correct_parallax(self):
             points = np.stack((height_lons, height_lats)).T
             h_interp = interpolate.griddata(points, heights, (lons, lats), method="linear")
 
-            new_lons, new_lats = get_parallax_corrected_lonlats(SAT_LON, SAT_LAT, SAT_ALT, lons, lats, h_interp)
+            new_lons, new_lats = get_parallax_corrected_lonlats(
+                SAT_LON, SAT_LAT, SAT_ALT, lons, lats, h_interp
+            )
 
             if updated_lats is None:
                 updated_lats = new_lats
@@ -98,7 +104,9 @@ def correct_parallax(self):
 
     def interp_at_grid(self, band: str, timestamp: datetime, target_grid: NDArray):
         self._load_previous_day()
-        assert (timestamp >= self.data["creation"]).any(), "Timestamp passed precedes all timestamps in the data"
+        assert (
+            timestamp >= self.data["creation"]
+        ).any(), "Timestamp passed precedes all timestamps in the data"
         closest_timestamp = self.data.loc[self.data["creation"] <= timestamp, "creation"].max()
         df = self.data[self.data["creation"] == closest_timestamp]
         column = f"{self.value}_{band}" if self.product == "ABI-L2-MCMIPF" else self.value
@@ -113,7 +121,10 @@ def interp_at_grid(self, band: str, timestamp: datetime, target_grid: NDArray):
         points = np.stack((x, y)).T
         shape = target_grid.shape[:2]
         interp_values = interpolate.griddata(
-            points, values, (target_grid[:, :, 1].flatten(), target_grid[:, :, 0].flatten()), method="linear"
+            points,
+            values,
+            (target_grid[:, :, 1].flatten(), target_grid[:, :, 0].flatten()),
+            method="linear",
         ).reshape(shape)
 
         return interp_values
diff --git a/pipelines/precipitation_model/impa/src/data/process/build_dataframe.py b/pipelines/precipitation_model/impa/src/data/process/build_dataframe.py
index 5beb243..16863be 100644
--- a/pipelines/precipitation_model/impa/src/data/process/build_dataframe.py
+++ b/pipelines/precipitation_model/impa/src/data/process/build_dataframe.py
@@ -12,7 +12,9 @@
 import numpy as np
 from tqdm import tqdm
 
-from pipelines.precipitation_model.impa.src.data.process.SatelliteData import SatelliteData
+from pipelines.precipitation_model.impa.src.data.process.SatelliteData import (
+    SatelliteData,
+)
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_warning
 
 HEAVY_RAIN_TRAIN_LOG_MEAN = 0.15839338
@@ -23,7 +25,14 @@
 
 
 def task_dt(
-    dt, ni: int, nj: int, sat_folder: str, grid_small: np.ndarray, grid_large: np.ndarray, value: str, band: str
+    dt,
+    ni: int,
+    nj: int,
+    sat_folder: str,
+    grid_small: np.ndarray,
+    grid_large: np.ndarray,
+    value: str,
+    band: str,
 ):
     date_str = dt.strftime("%Y-%m-%d")
     datetime_key = (dt).strftime("%Y%m%d/%H%M")
@@ -74,7 +83,9 @@ def build_dataframe(
         pass
 
     # Build dataframe for last 6 hours of data
-    dt_rounded = datetime.datetime(dt.year, dt.month, dt.day, dt.hour, dt.minute - dt.minute % timestep)
+    dt_rounded = datetime.datetime(
+        dt.year, dt.month, dt.day, dt.hour, dt.minute - dt.minute % timestep
+    )
     for i in range((60 * 6) // timestep):
         datetimes.append(dt_rounded - i * datetime.timedelta(minutes=timestep))
     full_datetimes = list(datetimes)
@@ -93,7 +104,8 @@ def build_dataframe(
             output_filepath.unlink()
         else:
             print_warning(
-                f"Warning: {output_filepath} already exists. Call with -o option to overwrite.", verbose=verbose
+                f"Warning: {output_filepath} already exists. Call with -o option to overwrite.",
+                verbose=verbose,
             )
             exit(0)
 
@@ -116,7 +128,8 @@ def build_dataframe(
         # these are the datetimes which we want to predict from
         relevant_datetimes = [datetimes[18], datetimes[0]]
         what.create_dataset(
-            "datetime_keys", data=np.asarray([dt.strftime("%Y%m%d/%H%M") for dt in relevant_datetimes], dtype="S")
+            "datetime_keys",
+            data=np.asarray([dt.strftime("%Y%m%d/%H%M") for dt in relevant_datetimes], dtype="S"),
         )
 
         step = num_workers
@@ -148,7 +161,9 @@ def build_dataframe(
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--verbose", "-v", action="store_true", help="If true, prints more information.")
+    parser.add_argument(
+        "--verbose", "-v", action="store_true", help="If true, prints more information."
+    )
     parser.add_argument(
         "--overwrite",
         "-o",
@@ -169,7 +184,13 @@ def build_dataframe(
         default=1,
         help="Number of processes for parallelization.",
     )
-    parser.add_argument("--location", "-loc", default="rio_de_janeiro", type=str, help="Location to build dataframe.")
+    parser.add_argument(
+        "--location",
+        "-loc",
+        default="rio_de_janeiro",
+        type=str,
+        help="Location to build dataframe.",
+    )
     parser.add_argument("--timestep", "-ts", default=10, type=int, help="Timestep in minutes.")
     parser.add_argument("--value", "-val", default="RRQPE", type=str, help="Satellite value.")
     parser.add_argument("--band", "-b", default="RRQPE", type=str, help="Satellite band.")
diff --git a/pipelines/precipitation_model/impa/src/eval/metrics/calc-metrics.py b/pipelines/precipitation_model/impa/src/eval/metrics/calc-metrics.py
index 4dba326..9961d1b 100644
--- a/pipelines/precipitation_model/impa/src/eval/metrics/calc-metrics.py
+++ b/pipelines/precipitation_model/impa/src/eval/metrics/calc-metrics.py
@@ -9,6 +9,7 @@
 import numpy as np
 import pandas as pd
 import tqdm
+
 from pipelines.precipitation_model.impa.src.eval.metrics.metrics import metrics_dict
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_warning
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import get_dataset_keys
diff --git a/pipelines/precipitation_model/impa/src/eval/update-real_time.py b/pipelines/precipitation_model/impa/src/eval/update-real_time.py
index 54d37d4..8f377c6 100644
--- a/pipelines/precipitation_model/impa/src/eval/update-real_time.py
+++ b/pipelines/precipitation_model/impa/src/eval/update-real_time.py
@@ -6,8 +6,13 @@
 import boto3
 from botocore import UNSIGNED
 from botocore.config import Config
-from pipelines.precipitation_model.impa.src.data.process.build_dataframe import build_dataframe
-from pipelines.precipitation_model.impa.src.data.process.process_satellite import process_satellite
+
+from pipelines.precipitation_model.impa.src.data.process.build_dataframe import (
+    build_dataframe,
+)
+from pipelines.precipitation_model.impa.src.data.process.process_satellite import (
+    process_satellite,
+)
 from pipelines.precipitation_model.impa.src.eval.predict_real_time import predict
 
 BUCKET_NAME = "noaa-goes16"
diff --git a/pipelines/precipitation_model/impa/src/eval/update-real_time_.py b/pipelines/precipitation_model/impa/src/eval/update-real_time_.py
index d92b504..24a5f76 100644
--- a/pipelines/precipitation_model/impa/src/eval/update-real_time_.py
+++ b/pipelines/precipitation_model/impa/src/eval/update-real_time_.py
@@ -10,8 +10,13 @@
 from botocore import UNSIGNED
 from botocore.config import Config
 from joblib import Parallel, delayed
-from pipelines.precipitation_model.impa.src.data.process.build_dataframe import build_dataframe
-from pipelines.precipitation_model.impa.src.data.process.process_satellite import process_satellite
+
+from pipelines.precipitation_model.impa.src.data.process.build_dataframe import (
+    build_dataframe,
+)
+from pipelines.precipitation_model.impa.src.data.process.process_satellite import (
+    process_satellite,
+)
 from pipelines.precipitation_model.impa.src.eval.predict_real_time import predict
 
 # from itertools import product
diff --git a/pipelines/precipitation_model/impa/src/eval/viz/plot-real_time.py b/pipelines/precipitation_model/impa/src/eval/viz/plot-real_time.py
index 9079a18..b09c5e8 100644
--- a/pipelines/precipitation_model/impa/src/eval/viz/plot-real_time.py
+++ b/pipelines/precipitation_model/impa/src/eval/viz/plot-real_time.py
@@ -10,6 +10,7 @@
 import numpy as np
 import pandas as pd
 import tqdm
+
 from pipelines.precipitation_model.impa.src.eval.metrics.metrics import metrics_dict
 from pipelines.precipitation_model.impa.src.utils.eval_utils import get_img
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_warning
diff --git a/pipelines/precipitation_model/impa/src/models/Evolution_Network/evolution_network.py b/pipelines/precipitation_model/impa/src/models/Evolution_Network/evolution_network.py
index e0b5f70..29911ea 100644
--- a/pipelines/precipitation_model/impa/src/models/Evolution_Network/evolution_network.py
+++ b/pipelines/precipitation_model/impa/src/models/Evolution_Network/evolution_network.py
@@ -1,7 +1,13 @@
 # -*- coding: utf-8 -*-
 import torch
 import torch.nn as nn
-from pipelines.precipitation_model.impa.src.models.Evolution_Network.model_parts import DoubleConv, Down, OutConv, Up
+
+from pipelines.precipitation_model.impa.src.models.Evolution_Network.model_parts import (
+    DoubleConv,
+    Down,
+    OutConv,
+    Up,
+)
 
 
 class Evolution_Encoder_Decoder(nn.Module):
diff --git a/pipelines/precipitation_model/impa/src/models/Evolution_Network/lightning.py b/pipelines/precipitation_model/impa/src/models/Evolution_Network/lightning.py
index 4956ebe..cf56d4a 100644
--- a/pipelines/precipitation_model/impa/src/models/Evolution_Network/lightning.py
+++ b/pipelines/precipitation_model/impa/src/models/Evolution_Network/lightning.py
@@ -1,7 +1,13 @@
 # -*- coding: utf-8 -*-
 import torch
-from pipelines.precipitation_model.impa.src.models.Evolution_Network.evolution_network import Evolution_Encoder_Decoder
-from pipelines.precipitation_model.impa.src.models.Evolution_Network.utils import make_grid, warp
+
+from pipelines.precipitation_model.impa.src.models.Evolution_Network.evolution_network import (
+    Evolution_Encoder_Decoder,
+)
+from pipelines.precipitation_model.impa.src.models.Evolution_Network.utils import (
+    make_grid,
+    warp,
+)
 from pipelines.precipitation_model.impa.src.models.lightning_module import LModule
 
 
diff --git a/pipelines/precipitation_model/impa/src/models/Metnet3/lightning.py b/pipelines/precipitation_model/impa/src/models/Metnet3/lightning.py
index 342ad12..07d736c 100644
--- a/pipelines/precipitation_model/impa/src/models/Metnet3/lightning.py
+++ b/pipelines/precipitation_model/impa/src/models/Metnet3/lightning.py
@@ -3,6 +3,8 @@
 
 import torch
 import torch.nn as nn
+from torch.nn import Sequential
+
 from pipelines.precipitation_model.impa.src.models.lightning_module import LModule
 from pipelines.precipitation_model.impa.src.models.Metnet3.Max_Vit.Max_Vit import MaxViT
 from pipelines.precipitation_model.impa.src.models.Metnet3.metnet3_pytorch import (
@@ -12,7 +14,6 @@
     ResnetBlocks,
     Upsample2x,
 )
-from torch.nn import Sequential
 
 
 class model(LModule):
diff --git a/pipelines/precipitation_model/impa/src/models/NowcastNet/lightning.py b/pipelines/precipitation_model/impa/src/models/NowcastNet/lightning.py
index 4145c4d..f377ad8 100644
--- a/pipelines/precipitation_model/impa/src/models/NowcastNet/lightning.py
+++ b/pipelines/precipitation_model/impa/src/models/NowcastNet/lightning.py
@@ -1,6 +1,10 @@
 # -*- coding: utf-8 -*-
 import torch
-from pipelines.precipitation_model.impa.src.models.gan.gan_unet.model import NowcasnetGenerator, TemporalDiscriminator
+
+from pipelines.precipitation_model.impa.src.models.gan.gan_unet.model import (
+    NowcasnetGenerator,
+    TemporalDiscriminator,
+)
 from pipelines.precipitation_model.impa.src.models.lightning_module import LModule
 
 
diff --git a/pipelines/precipitation_model/impa/src/models/UNET/lightning.py b/pipelines/precipitation_model/impa/src/models/UNET/lightning.py
index 8eeb335..de28661 100644
--- a/pipelines/precipitation_model/impa/src/models/UNET/lightning.py
+++ b/pipelines/precipitation_model/impa/src/models/UNET/lightning.py
@@ -1,7 +1,13 @@
 # -*- coding: utf-8 -*-
 import torch
+
 from pipelines.precipitation_model.impa.src.models.lightning_module import LModule
-from pipelines.precipitation_model.impa.src.models.UNET.unet_parts import DoubleConv, Down, OutConv, Up
+from pipelines.precipitation_model.impa.src.models.UNET.unet_parts import (
+    DoubleConv,
+    Down,
+    OutConv,
+    Up,
+)
 
 
 class model(LModule):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/attention.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/attention.py
index aa854a6..b695b10 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/attention.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/attention.py
@@ -4,11 +4,12 @@
 
 import torch
 import torch.nn.functional as F
-from torch import einsum, nn
 from einops import rearrange, repeat
+from torch import einsum, nn
 
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.utils import checkpoint
-
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.utils import (
+    checkpoint,
+)
 
 
 def exists(val):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoenc_v2.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoenc_v2.py
index a3a958f..d52393a 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoenc_v2.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoenc_v2.py
@@ -2,11 +2,17 @@
 import numpy as np
 import pytorch_lightning as pl
 import torch
+
 from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.distributions import (
     DiagonalGaussianDistribution,
 )
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.loss import LPIPSWithDiscriminator
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.modules import Decoder, Encoder
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.loss import (
+    LPIPSWithDiscriminator,
+)
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.modules import (
+    Decoder,
+    Encoder,
+)
 
 
 class AutoencoderKL(pl.LightningModule):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/loss.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/loss.py
index f7ac26b..09d4a74 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/loss.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/loss.py
@@ -3,11 +3,12 @@
 
 import torch
 import torch.nn.functional as F
+from torch import nn
+
 from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoencoder.discriminator import (
     NLayerDiscriminator,
     weights_init,
 )
-from torch import nn
 
 
 def hinge_d_loss(logits_real, logits_fake):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/modules.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/modules.py
index 891c041..ad12e94 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/modules.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/autoencoder/modules.py
@@ -8,7 +8,9 @@
 import torch.nn as nn
 
 # from ldm.util import instantiate_from_config
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.attention import LinearAttention
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.attention import (
+    LinearAttention,
+)
 
 
 class Encoder(nn.Module):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/ddim.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/ddim.py
index c73923c..a2b4b94 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/ddim.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/ddim.py
@@ -5,12 +5,13 @@
 
 import numpy as np
 import torch
+from tqdm import tqdm
+
 from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.utils import (
     make_ddim_sampling_parameters,
     make_ddim_timesteps,
     noise_like,
 )
-from tqdm import tqdm
 
 
 class DDIMSampler(object):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/lightning_v2.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/lightning_v2.py
index 33892ef..95c9e08 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/lightning_v2.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/lightning_v2.py
@@ -6,17 +6,26 @@
 import torch.nn.functional as F
 from einops import rearrange
 from pytorch_lightning import LightningModule
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoenc_v2 import AutoencoderKL
+from tqdm import tqdm
+
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.autoenc_v2 import (
+    AutoencoderKL,
+)
 
 # from pipelines.precipitation_model.impa.src.models.context_LDM_concat.autoencoder.autoenc_old import AutoencoderKL
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.ddim import DDIMSampler
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.ddim import (
+    DDIMSampler,
+)
 from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.model import (
     get_named_beta_schedule,
     linear_beta_schedule,
 )
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.openaimodel import UNetModel
-from pipelines.precipitation_model.impa.src.utils.data_utils import data_modification_options
-from tqdm import tqdm
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.openaimodel import (
+    UNetModel,
+)
+from pipelines.precipitation_model.impa.src.utils.data_utils import (
+    data_modification_options,
+)
 
 
 def extract(a, t, x_shape):
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/openaimodel.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/openaimodel.py
index ef9b3a4..577de2b 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/openaimodel.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/openaimodel.py
@@ -8,7 +8,10 @@
 import torch as th
 import torch.nn as nn
 import torch.nn.functional as F
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.attention import SpatialTransformer
+
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.attention import (
+    SpatialTransformer,
+)
 from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.utils import (
     avg_pool_nd,
     checkpoint,
diff --git a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/predict.py b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/predict.py
index 9fc14c9..ef6f900 100644
--- a/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/predict.py
+++ b/pipelines/precipitation_model/impa/src/models/context_LDM_concat_new/predict.py
@@ -5,8 +5,13 @@
 import pytorch_lightning as pl
 import torch
 from torch.utils.data import DataLoader
-from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import HDFDatasetLocations
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.lightning_v2 import Diffusion_Model
+
+from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import (
+    HDFDatasetLocations,
+)
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.lightning_v2 import (
+    Diffusion_Model,
+)
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import array_to_pred_hdf
 
diff --git a/pipelines/precipitation_model/impa/src/models/gan/gan_unet/model.py b/pipelines/precipitation_model/impa/src/models/gan/gan_unet/model.py
index eabde3b..07c5238 100644
--- a/pipelines/precipitation_model/impa/src/models/gan/gan_unet/model.py
+++ b/pipelines/precipitation_model/impa/src/models/gan/gan_unet/model.py
@@ -3,13 +3,26 @@
 
 import torch
 import torch.nn as nn
-from pipelines.precipitation_model.impa.src.models.gan.gan_unet.unet_parts import DoubleConv, Down, OutConv, S, Up, Up2
-from pipelines.precipitation_model.impa.src.models.gan.parts_gan import AvgPool, L2_Block, L3_Block, Noise_Projector
+from torch.nn.utils import spectral_norm
+
+from pipelines.precipitation_model.impa.src.models.gan.gan_unet.unet_parts import (
+    DoubleConv,
+    Down,
+    OutConv,
+    S,
+    Up,
+    Up2,
+)
+from pipelines.precipitation_model.impa.src.models.gan.parts_gan import (
+    AvgPool,
+    L2_Block,
+    L3_Block,
+    Noise_Projector,
+)
 from pipelines.precipitation_model.impa.src.models.nowcasting.layers.generation.generative_network import (
     Generative_Decoder,
     Generative_Encoder,
 )
-from torch.nn.utils import spectral_norm
 
 ni = 192  # size of image
 ndf = 32  # size of discriminator feature map
diff --git a/pipelines/precipitation_model/impa/src/models/lightning_module.py b/pipelines/precipitation_model/impa/src/models/lightning_module.py
index 0b21c98..df54a8e 100644
--- a/pipelines/precipitation_model/impa/src/models/lightning_module.py
+++ b/pipelines/precipitation_model/impa/src/models/lightning_module.py
@@ -3,7 +3,10 @@
 import torch
 from einops import rearrange
 from pytorch_lightning import LightningModule
-from pipelines.precipitation_model.impa.src.utils.data_utils import data_modification_options
+
+from pipelines.precipitation_model.impa.src.utils.data_utils import (
+    data_modification_options,
+)
 
 
 # define the LightningModule
diff --git a/pipelines/precipitation_model/impa/src/models/mamba/lightning.py b/pipelines/precipitation_model/impa/src/models/mamba/lightning.py
index 6f2fdac..deabd9c 100644
--- a/pipelines/precipitation_model/impa/src/models/mamba/lightning.py
+++ b/pipelines/precipitation_model/impa/src/models/mamba/lightning.py
@@ -3,8 +3,11 @@
 import torch
 from einops import rearrange
 from pytorch_lightning import LightningModule
+
 from pipelines.precipitation_model.impa.src.models.mamba.vmamba import VSSM
-from pipelines.precipitation_model.impa.src.utils.data_utils import data_modification_options
+from pipelines.precipitation_model.impa.src.utils.data_utils import (
+    data_modification_options,
+)
 
 
 class Vmamba_lightning(LightningModule):
diff --git a/pipelines/precipitation_model/impa/src/models/mamba/predict.py b/pipelines/precipitation_model/impa/src/models/mamba/predict.py
index 5abb79d..c501735 100644
--- a/pipelines/precipitation_model/impa/src/models/mamba/predict.py
+++ b/pipelines/precipitation_model/impa/src/models/mamba/predict.py
@@ -4,11 +4,16 @@
 
 import pytorch_lightning as pl
 import torch
-from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import HDFDatasetLocations
-from pipelines.precipitation_model.impa.src.models.mamba.lightning import Vmamba_lightning
+from torch.utils.data import DataLoader
+
+from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import (
+    HDFDatasetLocations,
+)
+from pipelines.precipitation_model.impa.src.models.mamba.lightning import (
+    Vmamba_lightning,
+)
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import array_to_pred_hdf
-from torch.utils.data import DataLoader
 
 MEAN = 0.08
 STD = 0.39
diff --git a/pipelines/precipitation_model/impa/src/models/nowcasting/layers/generation/generative_network.py b/pipelines/precipitation_model/impa/src/models/nowcasting/layers/generation/generative_network.py
index 6d9443c..7f527f9 100644
--- a/pipelines/precipitation_model/impa/src/models/nowcasting/layers/generation/generative_network.py
+++ b/pipelines/precipitation_model/impa/src/models/nowcasting/layers/generation/generative_network.py
@@ -3,8 +3,11 @@
 
 import torch.nn as nn
 import torch.nn.functional as F
+
 from pipelines.precipitation_model.impa.src.models.nowcasting.layers.evolution.module import *
-from pipelines.precipitation_model.impa.src.models.nowcasting.layers.generation.module import GenBlock
+from pipelines.precipitation_model.impa.src.models.nowcasting.layers.generation.module import (
+    GenBlock,
+)
 
 
 class Generative_Encoder(nn.Module):
diff --git a/pipelines/precipitation_model/impa/src/models/predict.py b/pipelines/precipitation_model/impa/src/models/predict.py
index 7e07a18..33ef557 100644
--- a/pipelines/precipitation_model/impa/src/models/predict.py
+++ b/pipelines/precipitation_model/impa/src/models/predict.py
@@ -7,13 +7,21 @@
 import pytorch_lightning as pl
 import torch
 from einops import rearrange
-from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import HDFDatasetLocations
-from pipelines.precipitation_model.impa.src.data.PredHDFDatasetLocations import PredHDFDatasetLocations
+from torch.utils.data import DataLoader
+
+from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import (
+    HDFDatasetLocations,
+)
+from pipelines.precipitation_model.impa.src.data.PredHDFDatasetLocations import (
+    PredHDFDatasetLocations,
+)
 from pipelines.precipitation_model.impa.src.utils.data_utils import sat_dataframe
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import array_to_pred_hdf
-from pipelines.precipitation_model.impa.src.utils.models_utils import get_ds, options_pretrained
-from torch.utils.data import DataLoader
+from pipelines.precipitation_model.impa.src.utils.models_utils import (
+    get_ds,
+    options_pretrained,
+)
 
 MEAN_LOG_SAT = 0.08
 STD_LOG_SAT = 0.39
diff --git a/pipelines/precipitation_model/impa/src/models/pysteps_LK/predict.py b/pipelines/precipitation_model/impa/src/models/pysteps_LK/predict.py
index 1b03482..4205a39 100644
--- a/pipelines/precipitation_model/impa/src/models/pysteps_LK/predict.py
+++ b/pipelines/precipitation_model/impa/src/models/pysteps_LK/predict.py
@@ -7,7 +7,10 @@
 from joblib import Parallel, delayed
 from pysteps import motion, nowcasts
 from pysteps.utils import conversion, transformation
-from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import HDFDatasetLocations
+
+from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import (
+    HDFDatasetLocations,
+)
 from pipelines.precipitation_model.impa.src.utils.general_utils import print_ok
 from pipelines.precipitation_model.impa.src.utils.hdf_utils import array_to_pred_hdf
 
diff --git a/pipelines/precipitation_model/impa/src/utils/eval_utils.py b/pipelines/precipitation_model/impa/src/utils/eval_utils.py
index 9be85a8..38ca89d 100644
--- a/pipelines/precipitation_model/impa/src/utils/eval_utils.py
+++ b/pipelines/precipitation_model/impa/src/utils/eval_utils.py
@@ -8,11 +8,20 @@
 import pandas as pd
 import plotly.graph_objects as go
 from PIL import Image
+
 from pipelines.precipitation_model.impa.src.eval.metrics.metrics import metrics_dict
-from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.predict import main as latent_diffusion_predict
-from pipelines.precipitation_model.impa.src.models.mamba.predict import main as mamba_predict
-from pipelines.precipitation_model.impa.src.models.predict import main as general_predict
-from pipelines.precipitation_model.impa.src.models.pysteps_LK.predict import main as pysteps_predict
+from pipelines.precipitation_model.impa.src.models.context_LDM_concat_new.predict import (
+    main as latent_diffusion_predict,
+)
+from pipelines.precipitation_model.impa.src.models.mamba.predict import (
+    main as mamba_predict,
+)
+from pipelines.precipitation_model.impa.src.models.predict import (
+    main as general_predict,
+)
+from pipelines.precipitation_model.impa.src.models.pysteps_LK.predict import (
+    main as pysteps_predict,
+)
 
 MAP_CENTER = {"lat": -22.914550816555533, "lon": -43.502443050594596}
 ZOOM = 8
diff --git a/pipelines/precipitation_model/impa/src/utils/models_utils.py b/pipelines/precipitation_model/impa/src/utils/models_utils.py
index 2fa09ec..ac86fa7 100644
--- a/pipelines/precipitation_model/impa/src/utils/models_utils.py
+++ b/pipelines/precipitation_model/impa/src/utils/models_utils.py
@@ -1,9 +1,15 @@
 # -*- coding: utf-8 -*-
 from pipelines.precipitation_model.impa.src.data.HDFDataset2 import HDFDataset2
-from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import HDFDatasetLocations
-from pipelines.precipitation_model.impa.src.data.HDFDatasetMerged import HDFDatasetMerged
+from pipelines.precipitation_model.impa.src.data.HDFDatasetLocations import (
+    HDFDatasetLocations,
+)
+from pipelines.precipitation_model.impa.src.data.HDFDatasetMerged import (
+    HDFDatasetMerged,
+)
 from pipelines.precipitation_model.impa.src.data.PredHDFDataset2 import PredHDFDataset2
-from pipelines.precipitation_model.impa.src.data.PredHDFDatasetLocations import PredHDFDatasetLocations
+from pipelines.precipitation_model.impa.src.data.PredHDFDatasetLocations import (
+    PredHDFDatasetLocations,
+)
 
 extension_dicts = {
     "context_LDM_VAE": ".pt",
diff --git a/pipelines/precipitation_model/impa/tasks.py b/pipelines/precipitation_model/impa/tasks.py
index 4f7c8fe..80971ba 100644
--- a/pipelines/precipitation_model/impa/tasks.py
+++ b/pipelines/precipitation_model/impa/tasks.py
@@ -11,10 +11,14 @@
 from botocore.config import Config
 from prefect import task
 from prefeitura_rio.pipelines_utils.logging import log
-from pipelines.precipitation_model.impa.src.data.process.build_dataframe import build_dataframe
-from pipelines.precipitation_model.impa.src.data.process.process_satellite import process_satellite
-from pipelines.precipitation_model.impa.src.eval.predict_real_time import predict
 
+from pipelines.precipitation_model.impa.src.data.process.build_dataframe import (
+    build_dataframe,
+)
+from pipelines.precipitation_model.impa.src.data.process.process_satellite import (
+    process_satellite,
+)
+from pipelines.precipitation_model.impa.src.eval.predict_real_time import predict
 from pipelines.precipitation_model.impa.utils import download_file_from_s3
 
 
@@ -85,7 +89,7 @@ def process_data(year, day_of_year, num_workers, dt):
         num_workers (int): The number of workers to use for parallel processing.
         dt (datetime.datetime): The datetime object representing the date to process.
 
-    This function logs the processing activity, processes satellite data for specified 
+    This function logs the processing activity, processes satellite data for specified
     products using `process_satellite`, and then builds a dataframe with `build_dataframe`.
     """
     log("Processing satellite data...")