Support recent versions of black and lightning

tests/test_cli.py

@@ -64,7 +64,6 @@ def test_shared_cli_options(mocker, minimum_valid_train, minimum_valid_predict):
     mocker.patch("zamba.cli.ModelManager.predict", pred_mock)
 
     for command in [minimum_valid_train, minimum_valid_predict]:
-
         # check default model is time distributed one
         result = runner.invoke(app, command)
         assert result.exit_code == 0

tests/test_load_video_frames.py

@@ -65,7 +65,6 @@ def assert_megadetector_total_or_none(original_video_metadata, video_shape, **kw
 
 
 def assert_no_frames_or_correct_shape(original_video_metadata, video_shape, **kwargs):
-
     return (video_shape["frames"] == 0) or (
         (video_shape["height"] == kwargs["frame_selection_height"])
         and (video_shape["width"] == kwargs["frame_selection_width"])

zamba/cli.py

@@ -600,5 +600,4 @@ def depth(
 
 
 if __name__ == "__main__":
-
     app()

zamba/models/config.py

@@ -148,7 +148,6 @@ def check_files_exist_and_load(
 
     bad_load = []
     if not skip_load_validation:
-
         logger.info(
             "Checking that all videos can be loaded. If you're very confident all your videos can be loaded, you can skip this with `skip_load_validation`, but it's not recommended."
         )
@@ -503,7 +502,6 @@ def validate_filepaths_and_labels(cls, values):
 
         # validate split column has no partial nulls or invalid values
         if "split" in labels.columns:
-
             # if split is entirely null, warn, drop column, and generate splits automatically
             if labels.split.isnull().all():
                 logger.warning(
@@ -559,7 +557,6 @@ def validate_provided_species_and_use_default_model_labels(cls, values):
         )
 
         if not provided_species.issubset(model_species):
-
             # if labels are not a subset, user cannot set use_default_model_labels to True
             if values["use_default_model_labels"]:
                 raise ValueError(
@@ -677,7 +674,6 @@ def make_split(labels, values):
             species_df = labels[labels[c] > 0]
 
             if len(species_df):
-
                 # within each species, seed splits by putting one video in each set and then allocate videos based on split proportions
                 labels.loc[species_df.index, "split"] = expected_splits + random.choices(
                     list(values["split_proportions"].keys()),

zamba/models/depth_estimation/depth_manager.py

@@ -36,7 +36,6 @@ def depth_transforms(size):
 
 class DepthDataset(torch.utils.data.Dataset):
     def __init__(self, filepaths):
-
         # these are hardcoded because they depend on the trained model weights used for inference
         self.height = 270
         self.width = 480
@@ -55,7 +54,6 @@ def __init__(self, filepaths):
 
         logger.info(f"Running object detection on {len(filepaths)} videos.")
         for video_filepath in tqdm(filepaths):
-
             # get video array at 1 fps, use full size for detecting objects
             logger.debug(f"Loading video: {video_filepath}")
             try:
@@ -73,7 +71,6 @@ def __init__(self, filepaths):
 
             # iterate over frames
             for frame_idx, (detections, scores) in enumerate(detections_per_frame):
-
                 # if anything is detected in the frame, save out relevant frames
                 if len(detections) > 0:
                     logger.debug(f"{len(detections)} detection(s) found at second {frame_idx}.")
@@ -234,7 +231,10 @@ def predict(self, filepaths):
                     for d, vid, t in zip(distance.cpu().numpy(), filepath, time):
                         predictions.append((vid, t, d))
 
-        predictions = pd.DataFrame(predictions, columns=["filepath", "time", "distance"],).round(
+        predictions = pd.DataFrame(
+            predictions,
+            columns=["filepath", "time", "distance"],
+        ).round(
             {"distance": 1}
         )  # round to useful number of decimal places
 

zamba/models/efficientnet_models.py

@@ -19,7 +19,6 @@ class TimeDistributedEfficientNet(ZambaVideoClassificationLightningModule):
     def __init__(
         self, num_frames=16, finetune_from: Optional[Union[os.PathLike, str]] = None, **kwargs
     ):
-
         super().__init__(**kwargs)
 
         if finetune_from is None:

zamba/models/model_manager.py

@@ -390,7 +390,6 @@ def predict_model(
     }
 
     if predict_config.save is not False:
-
         config_path = predict_config.save_dir / "predict_configuration.yaml"
         logger.info(f"Writing out full configuration to {config_path}.")
         with config_path.open("w") as fp:
@@ -415,7 +414,6 @@ def predict_model(
         df = df.round(5)
 
     if predict_config.save is not False:
-
         preds_path = predict_config.save_dir / "zamba_predictions.csv"
         logger.info(f"Saving out predictions to {preds_path}.")
         with preds_path.open("w") as fp:

zamba/pytorch_lightning/utils.py

@@ -6,6 +6,7 @@
 
 import numpy as np
 import pandas as pd
+import pytorch_lightning as pl
 from pytorch_lightning import LightningDataModule, LightningModule
 from sklearn.metrics import f1_score, top_k_accuracy_score, accuracy_score
 import torch
@@ -273,9 +274,17 @@ def configure_optimizers(self):
             }
 
     def to_disk(self, path: os.PathLike):
+        """Save out model weights to a checkpoint file on disk.
+
+        Note: this does not include callbacks, optimizer_states, or lr_schedulers.
+        To include those, use `Trainer.save_checkpoint()` instead.
+        """
+
         checkpoint = {
             "state_dict": self.state_dict(),
             "hyper_parameters": self.hparams,
+            "global_step": self.global_step,
+            "pytorch-lightning_version": pl.__version__,
         }
         torch.save(checkpoint, path)