Remove detect_anomaly and quiet

gaussian_splatting/training.py

@@ -21,17 +21,15 @@ def __init__(
         saving_iterations=None,
         checkpoint_iterations=None,
         checkpoint_path=None,
-        quiet=False,
-        detect_anomaly=False,
     ):
         self._resolution = resolution
 
         if testing_iterations is None:
-            testing_iterations = [7_000, 30_000]
+            testing_iterations = [7000, 30000]
         self._testing_iterations = testing_iterations
 
         if saving_iterations is None:
-            saving_iterations = [7_000, 30_000]
+            saving_iterations = [7000, 30000]
         self._saving_iterations = saving_iterations
 
         if checkpoint_iterations is None:
@@ -40,10 +38,6 @@ def __init__(
 
         self._checkpoint_path = checkpoint_path
 
-        safe_state(quiet)
-
-        torch.autograd.set_detect_anomaly(detect_anomaly)
-
     def run(
         self,
         dataset,
@@ -220,4 +214,4 @@ def training_report(
 
         torch.cuda.empty_cache()
 
-    print("\nTraining complete.")
+        print("\nTraining complete.")

gaussian_splatting/utils/general.py

@@ -120,7 +120,7 @@ def build_scaling_rotation(s, r):
     return L
 
 
-def safe_state(silent):
+def safe_state(silent=True, seed=0):
     old_f = sys.stdout
 
     class F:
@@ -146,7 +146,7 @@ def flush(self):
 
     sys.stdout = F(silent)
 
-    random.seed(0)
-    np.random.seed(0)
-    torch.manual_seed(0)
+    random.seed(seed)
+    np.random.seed(seed)
+    torch.manual_seed(seed)
     torch.cuda.set_device(torch.device("cuda:0"))

scripts/full_eval.py

@@ -37,7 +37,7 @@
     args = parser.parse_args()
 
 if not args.skip_training:
-    common_args = " --quiet --eval --test_iterations -1 "
+    common_args = " --eval --test_iterations -1 "
     for scene in mipnerf360_outdoor_scenes:
         source = args.mipnerf360 + "/" + scene
         os.system(
@@ -94,7 +94,7 @@
     for scene in deep_blending_scenes:
         all_sources.append(args.deepblending + "/" + scene)
 
-    common_args = " --quiet --eval --skip_train"
+    common_args = " --eval --skip_train"
     for scene, source in zip(all_scenes, all_sources):
         os.system(
             "python render.py --iteration 7000 -s "

scripts/render.py

@@ -84,11 +84,9 @@ def render_sets(
     parser.add_argument("--iteration", default=-1, type=int)
     parser.add_argument("--skip_train", action="store_true")
     parser.add_argument("--skip_test", action="store_true")
-    parser.add_argument("--quiet", action="store_true")
     args = get_combined_args(parser)
     print("Rendering " + args.model_path)
 
-    # Initialize system state (RNG)
-    safe_state(args.quiet)
+    safe_state()
 
     render_sets(model.extract(args), args.iteration, args.skip_train, args.skip_test)

scripts/train.py

@@ -19,14 +19,12 @@
     parser = ArgumentParser(description="Training script parameters")
     lp = ModelParams(parser)
     op = OptimizationParams(parser)
-    parser.add_argument("--detect_anomaly", action="store_true", default=False)
     parser.add_argument(
         "--test_iterations", nargs="+", type=int, default=[7_000, 30_000]
     )
     parser.add_argument(
         "--save_iterations", nargs="+", type=int, default=[7_000, 30_000]
     )
-    parser.add_argument("--quiet", action="store_true")
     parser.add_argument("--checkpoint_iterations", nargs="+", type=int, default=[])
     parser.add_argument("--checkpoint_path", type=str, default=None)
     parser.add_argument("--resolution", default=-1, type=int)
@@ -39,8 +37,6 @@
         saving_iterations=args.save_iterations,
         checkpoint_iterations=args.checkpoint_iterations,
         checkpoint_path=args.checkpoint_path,
-        quiet=args.quiet,
-        detect_anomaly=args.detect_anomaly,
     )
     trainer.run(
         dataset=lp.extract(args),