add plots for validation score during training. Fixed setting model t…

…o eval and train modes

neural_networks/create_grid_search.py

@@ -0,0 +1,11 @@
+with open("parameters.txt", "w") as f:
+    i = 0
+    for model_name in ["dinov2_vitl14_reg", "dinov2_vitb14_reg", "efficientnet_v2_l"]:
+        for lr, batch_size in [(1e-4, 32), (1e-4, 64), (2e-4, 64), (5e-5, 32)]:
+            for dropout_p in [0.1, 0.15, 0.2, 0.25]:
+                for label_smoothing in [0.1, 0.2, 0.3]:
+                    i += 1
+                    cmd = f"{model_name} {lr} 1 {dropout_p} {batch_size} {label_smoothing} 0 0 16 16 {560 if 'dinov2' in model_name else 0} stack 0\n"
+                    print(cmd)
+                    f.writelines(cmd)
+print(i)

neural_networks/parameters.txt

@@ -1,79 +1,144 @@
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.2 0 1 16 16 560
-
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.2 0 1 16 16 560
-
-dinov2_vitl14_reg 1e-05 1 0.25 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.25 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.25 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.25 32 0.2 0 1 16 16 560
-
-dinov2_vitl14_reg 1e-05 1 0.1 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.1 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.1 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 1e-05 1 0.1 32 0.2 0 1 16 16 560
-
-dinov2_vitl14_reg 5e-05 1 0.25 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.25 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.25 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.25 32 0.2 0 1 16 16 560
-
-dinov2_vitl14_reg 5e-05 1 0.1 32 0.1 0 0 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.1 32 0.1 1 1 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.1 32 0.2 1 0 16 16 560
-dinov2_vitl14_reg 5e-05 1 0.1 32 0.2 0 1 16 16 560
-
-efficientnet_v2_l 1e-04 1 0.25 32 0.2 0 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.25 32 0.2 1 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.25 32 0.1 1 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.25 32 0.1 0 0 16 16 0
-
-efficientnet_v2_l 1e-04 1 0.1 32 0.2 0 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.1 32 0.2 1 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.1 32 0.1 1 0 16 16 0
-efficientnet_v2_l 1e-04 1 0.1 32 0.1 0 0 16 16 0
-
-efficientnet_v2_l 5e-05 1 0.25 32 0.2 0 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.25 32 0.2 1 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.25 32 0.1 1 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.25 32 0.1 0 0 16 16 0
-
-efficientnet_v2_l 5e-05 1 0.1 32 0.2 0 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.1 32 0.2 1 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.1 32 0.1 1 0 16 16 0
-efficientnet_v2_l 5e-05 1 0.1 32 0.1 0 0 16 16 0
-
-efficientnet_v2_l 1e-05 1 0.25 32 0.2 0 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.25 32 0.2 1 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.25 32 0.1 1 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.25 32 0.1 0 0 16 16 0
-
-efficientnet_v2_l 1e-05 1 0.1 32 0.2 0 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.1 32 0.2 1 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.1 32 0.1 1 0 16 16 0
-efficientnet_v2_l 1e-05 1 0.1 32 0.1 0 0 16 16 0
-
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 0 16 16 560 conv 0
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 0 16 16 560 conv 0
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 0 16 16 560 conv 1
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 0 16 16 560 conv 1
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 0 16 16 560 stack 0
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 0 16 16 560 stack 0
-efficientnet_v2_l 1e-04 1 0.1 32 0.2 0 0 16 16 0 stack 0
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 0 16 16 560 stack 1
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 0 16 16 560 stack 1
-efficientnet_v2_l 1e-04 1 0.1 32 0.2 0 0 16 16 0 stack 1
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 1 16 16 560 conv 0
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 1 16 16 560 conv 0
-dinov2_vitl14_reg 1e-04 1 0.25 32 0.1 0 1 16 16 560 conv 1
-dinov2_vitl14_reg 1e-04 1 0.1 32 0.1 0 1 16 16 560 conv 1
-
-
-
-
-
+dinov2_vitl14_reg 0.0001 1 0.1 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.1 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.1 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.1 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.1 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.1 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.15 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.2 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0001 1 0.25 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.1 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.1 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.1 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.15 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.15 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.15 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.2 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.2 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.2 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.25 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.25 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 0.0002 1 0.25 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.1 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.1 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.1 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.15 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.15 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.15 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.2 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.2 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.2 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.25 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.25 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitl14_reg 5e-05 1 0.25 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.1 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.15 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.2 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0001 1 0.25 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.1 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.1 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.1 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.15 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.15 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.15 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.2 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.2 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.2 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.25 64 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.25 64 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 0.0002 1 0.25 64 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.1 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.1 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.1 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.15 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.15 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.15 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.2 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.2 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.2 32 0.3 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.25 32 0.1 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.25 32 0.2 0 0 16 16 560 stack 0
+dinov2_vitb14_reg 5e-05 1 0.25 32 0.3 0 0 16 16 560 stack 0
+efficientnet_v2_l 0.0001 1 0.1 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.1 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.1 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.1 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.1 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.1 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.15 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.2 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0001 1 0.25 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.1 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.1 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.1 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.15 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.15 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.15 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.2 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.2 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.2 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.25 64 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.25 64 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 0.0002 1 0.25 64 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.1 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.1 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.1 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.15 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.15 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.15 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.2 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.2 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.2 32 0.3 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.25 32 0.1 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.25 32 0.2 0 0 16 16 0 stack 0
+efficientnet_v2_l 5e-05 1 0.25 32 0.3 0 0 16 16 0 stack 0

neural_networks/plots/score_plots.py

@@ -0,0 +1,61 @@
+import matplotlib.pyplot as plt
+from tensorboard.backend.event_processing import event_accumulator
+
+
+def plot_tensorboard_timeseries(event_file, tag, epochs=120):
+    # Load TensorBoard event file
+    ea = event_accumulator.EventAccumulator(event_file)
+    ea.Reload()
+
+    # Extract the time series data
+    if tag not in ea.Tags()["scalars"]:
+        raise ValueError(f"Tag '{tag}' not found in the event file.")
+
+    events = ea.Scalars(tag)
+    steps = [e.step / 54 for e in events][:epochs]
+    values = [e.value for e in events][:epochs]
+    total_time = events[epochs - 1].wall_time - events[0].wall_time
+    return steps, values, total_time
+
+
+def plot_scores(stepsLists, valueLists, wall_times, models):
+
+    # Plot the time series
+    plt.figure(figsize=(10, 5))
+    for model_steps, value_values, wall_time, model in zip(
+        stepsLists, valueLists, wall_times, models
+    ):
+        print(
+            f"{model} wall time was {wall_time:.0f} seconds ({wall_time/3600:.2f} hours)"
+        )
+        # print(model_steps, value_values, model)
+        plt.plot(model_steps, value_values, label=model)
+    plt.xlabel("Epoch")
+    plt.ylabel("Area under PR-curve")
+    plt.legend()
+    plt.grid(True)
+    plt.savefig("au_pr_score.png")
+
+
+# Usage example
+event_file = "event_files/events.out.tfevents.1730291782.gcn30.local.snellius.surf.nl.2180618.0"  # Update with your TensorBoard event file path
+event_file_dict = {
+    "DinoV2": "event_files/events.out.tfevents.1730291782.gcn30.local.snellius.surf.nl.2180618.0",
+    "EfficientNet": "event_files/events.out.tfevents.1730302269.gcn30.local.snellius.surf.nl.2222324.0",
+}
+
+# Extract data for each model
+stepsLists, valuesLists, wall_times, models = zip(
+    *[
+        (
+            *plot_tensorboard_timeseries(
+                event_file_dict[model], "au_pr_curve/validation", epochs=120
+            ),
+            model,
+        )
+        for model in event_file_dict.keys()
+    ]
+)
+
+
+plot_scores(stepsLists, valuesLists, wall_times, models)

neural_networks/pre_processing_for_ml.py

@@ -108,6 +108,7 @@ def process_fits(fits_path):
         for fits_path in (root_dir / (cls + mode)).glob("*.fits")
     )
 
+
 class FitsDataset(Dataset):
     def __init__(self, root_dir, mode="train"):
         """
@@ -168,7 +169,7 @@ def __init__(self, root_dir, mode="train"):
         # print(f'{mode}: using the following sources: {sources}')
 
     def compute_statistics(self, normalize):
-        cache = Memory(location=self.root_dir / '_cache')
+        cache = Memory(location=self.root_dir / "_cache")
         cached_compute = cache.cache(FitsDataset._compute_statistics)
         self.mean, self.std = cached_compute(self, normalize)
         return self.mean, self.std

neural_networks/train_nn.job

@@ -1,6 +1,6 @@
 #!/bin/bash
 #SBATCH --job-name=cortex_grid_search
-#SBATCH -p gpu_h100
+#SBATCH -p gpu
 #SBATCH -t 08:00:00
 #SBATCH --gpus 1
 #SBATCH --output=out/multi_cortex%A_%a.out
@@ -52,8 +52,12 @@ else
     EPOCHS="120"
 fi
 
+DATA_INPUT_PATH="/scratch-shared/CORTEX/"
+# find $DATA_INPUT_PATH -name '*npz' | xargs -n 1 -P 18 -i rsync -R {} '/dev/shm/'
+
 DATA_TRAINDATA_PATH="/scratch-shared/CORTEX/public.spider.surfsara.nl/lofarvwf/jdejong/CORTEX/calibrator_selection_robertjan/cnn_data/"
 
+
 # Execute your Python script with the given parameters
 echo  $DATA_TRAINDATA_PATH --model $model --lr $lr --normalize $normalize --dropout_p $dropout_p --batch_size $batch_size --log_path grid_search_2 --label_smoothing $label_smoothing --rank $rank --resize $resize --alpha $alpha  $LORA_ARG $STOCHASTIC_SMOOTHING -d --epochs $EPOCHS --lift $lift $FLIP_AUGMENTATIONS 
 python train_nn.py $DATA_TRAINDATA_PATH --model $model --lr $lr --normalize $normalize --dropout_p $dropout_p --batch_size $batch_size --log_path grid_search_2 --label_smoothing $label_smoothing --rank $rank --resize $resize --alpha $alpha  $LORA_ARG $STOCHASTIC_SMOOTHING -d --epochs $EPOCHS --lift $lift $FLIP_AUGMENTATIONS 

neural_networks/train_nn.py

@@ -113,6 +113,7 @@ def init_first_conv(conv):
 
 def get_classifier(dropout_p: float, n_features: int, num_target_classes: int):
     assert 0 <= dropout_p <= 1
+    print(n_features)
 
     classifier = nn.Sequential(
         nn.Flatten(),
@@ -800,7 +801,7 @@ def get_argparser():
         "--normalize",
         type=int,
         help="Whether to do normalization",
-        default=0,
+        default=1,
         choices=[0, 1, 2],
     )
     parser.add_argument(