Disable pesky mean warnings

spanet/network/jet_reconstruction/jet_reconstruction_validation.py

@@ -1,4 +1,5 @@
 from typing import Dict, Callable
+import warnings
 
 import numpy as np
 import torch
@@ -62,13 +63,16 @@ def compute_metrics(self, jet_predictions, particle_scores, stacked_targets, sta
  particle_accuracies = particle_accuracies.max(0)
 
  # Create the logging dictionaries
- metrics = {f"jet/accuracy_{i}_of_{j}": (jet_accuracies[num_particles == j] >= i).mean()
- for j in range(1, num_targets + 1)
- for i in range(1, j + 1)}
-
- metrics.update({f"particle/accuracy_{i}_of_{j}": (particle_accuracies[num_particles == j] >= i).mean()
- for j in range(1, num_targets + 1)
- for i in range(1, j + 1)})
+ with warnings.catch_warnings():
+ warnings.simplefilter("ignore", category=RuntimeWarning)
+
+ metrics = {f"jet/accuracy_{i}_of_{j}": (jet_accuracies[num_particles == j] >= i).mean()
+ for j in range(1, num_targets + 1)
+ for i in range(1, j + 1)}
+
+ metrics.update({f"particle/accuracy_{i}_of_{j}": (particle_accuracies[num_particles == j] >= i).mean()
+ for j in range(1, num_targets + 1)
+ for i in range(1, j + 1)})
 
  particle_scores = particle_scores.ravel()
  particle_targets = permuted_masks.ravel()