Fix linter issues

Author: DubiousCactus

.github/workflows/python-app.yml

@@ -37,10 +37,10 @@ jobs:
         flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
     - name: Type check with mypy (training script)
       run: |
-        mypy train.py +experiment=exp_a run.epochs=2 run.viz_every=0
+        mypy train.py
     - name: Type check with mypy (test script)
       run: |
-        mypy test.py +experiment=exp_a run.viz_every=0
+        mypy test.py
     - name: Test with pytest
       run: |
         pytest

launch_experiment.py

@@ -18,7 +18,7 @@
 from hydra_zen import just
 from hydra_zen.typing import Partial
 
-import conf.experiment  # Must import the config to add all components to the store!
+import conf.experiment  # Must import the config to add all components to the store! # noqa
 from conf import project as project_conf
 from model import TransparentDataParallel
 from src.base_trainer import BaseTrainer

src/base_tester.py

@@ -79,7 +79,7 @@ def test(self, visualize_every: int = 0, **kwargs):
             visualize_every (int, optional): Visualize the model predictions every n batches.
             Defaults to 0 (no visualization).
         """
-        test_loss, test_loss_components = MeanMetric(), defaultdict(MeanMetric)
+        test_loss, test_metrics = MeanMetric(), defaultdict(MeanMetric)
         self._model.eval()
         self._pbar.reset()
         self._pbar.set_description("Testing")
@@ -92,7 +92,7 @@ def test(self, visualize_every: int = 0, **kwargs):
             loss, metrics = self._test_iteration(batch)
             test_loss.update(loss.item())
             for k, v in metrics.items():
-                metrics[k].update(v.item())
+                test_metrics[k].update(v.item())
             update_pbar_str(
                 self._pbar,
                 f"Testing [loss={test_loss.compute():.4f}]",
@@ -107,7 +107,7 @@ def test(self, visualize_every: int = 0, **kwargs):
         print("=" * 81)
         print("==" + " " * 31 + " Test results " + " " * 31 + "==")
         print("=" * 81)
-        for k, v in metrics.items():
+        for k, v in test_metrics.items():
             print(f"\t -> {k}: {v.compute().item():.2f}")
         print(f"\t -> Average loss: {test_loss:.4f}")
         print("_" * 81)

src/base_trainer.py

@@ -323,7 +323,7 @@ def _plot(self, epoch: int, train_losses: List[float], val_losses: List[float]):
         plt.theme("dark")
         plt.xlabel("Epoch")
         if project_conf.LOG_SCALE_PLOT:
-            if any(l <= 0 for l in train_losses + val_losses):
+            if any(loss_val <= 0 for loss_val in train_losses + val_losses):
                 raise ValueError(
                     "Cannot plot on a log scale if there are non-positive losses."
                 )

test.py

@@ -8,7 +8,7 @@
 
 from hydra_zen import store, zen
 
-import conf.experiment  # Must import the config to add all components to the store!
+import conf.experiment  # Must import the config to add all components to the store! # noqa
 from conf import project as project_conf
 from launch_experiment import launch_experiment
 from utils import seed_everything

train.py

@@ -8,7 +8,7 @@
 
 from hydra_zen import store, zen
 
-import conf.experiment  # Must import the config to add all components to the store!
+import conf.experiment  # Must import the config to add all components to the store! # noqa
 from conf import project as project_conf
 from launch_experiment import launch_experiment
 from utils import seed_everything

utils/__init__.py

@@ -102,7 +102,7 @@ def blink_pbar(i: int, pbar: tqdm.tqdm, n: int) -> None:
 
 @contextmanager
 def colorize_prints(ansii_code: Union[int, str]):
-    if type(ansii_code) == str:
+    if type(ansii_code) is str:
         ansii_code = project_conf.ANSI_COLORS[ansii_code]
     print(f"\033[{ansii_code}m", end="")
     try:
@@ -180,7 +180,7 @@ def wrapper(*args, **kwargs):
                 )
 
                 if reload.lower() not in ("l", "", "r"):
-                    print(f"[!] Aborting")
+                    print("[!] Aborting")
                     # TODO: Why can't I just raise the exception? It's weird but it gets caught by
                     # the wrapper a few times until it finally gets raised.
                     sys.exit(1)
@@ -196,7 +196,7 @@ def wrapper(*args, **kwargs):
                         cfg=IPython.terminal.embed.load_default_config(),
                         banner1=colorize(
                             f"[*] Dropping into an IPython shell to inspect {callable} "
-                            + f"with the locals as they were at the time of the exception "
+                            + "with the locals as they were at the time of the exception "
                             + f"thrown at line {frame.f_lineno} of {frame.f_code.co_filename}."
                             + "\n============================== TIPS =============================="
                             + "\n -> Use '%whos' to list variables in the current scope."

utils/anim.py

@@ -76,9 +76,11 @@ def __init__(
         super().__init__()
         try:
             import scenepic as sp
-        except:
+        except ImportError:
             raise Exception(
-                "scenepic not installed. Some visualization functions will not work. (I know it's not available on Apple Silicon :("
+                "scenepic not installed. "
+                + "Some visualization functions will not work. "
+                + "(I know it's not available on Apple Silicon :("
             )
         pv.start_xvfb()
         self.scene = sp.Scene()

utils/helpers.py

@@ -45,7 +45,9 @@ def __call__(
         ):  # Either val_loss or min_val_metric
             ckpt_path = osp.join(
                 HydraConfig.get().runtime.output_dir,
-                f"epoch_{epoch:03d}_{minimize_metric if minimize_metric in metrics.keys() else 'val-loss'}_{metrics.get(minimize_metric, val_loss):06f}.ckpt",
+                f"epoch_{epoch:03d}_"
+                + f"{minimize_metric if minimize_metric in metrics.keys() else 'val-loss'}"
+                + f"_{metrics.get(minimize_metric, val_loss):06f}.ckpt",
             )
             self._save_if_best_model(
                 metrics.get(minimize_metric, val_loss), ckpt_path