Add files via upload

In der config Datei kann man nun einstellen, ob während des Training mit leaky relu, bei einer Performance von 100% auf relu geswitched wird (d.h. leaky relu mit slope = 0.0). In der cnn_trainin.py musste ich beim Lesen und Laden der config.json aufgrund eines komischen Errors beim Ausführen der .sh-file was ändern.

Classic_contour_net_shallow/classic_loop.sh

@@ -2,7 +2,7 @@ Directory="/home/kk/Documents/Semester4/code/Classic_contour_net_shallow"
 Priority="-500"
 echo $Directory
 mkdir $Directory/argh_log_classic
-for out_channels_idx in {0..0}; do
+for out_channels_idx in {0..6}; do
   for kernel_size_idx in {0..0}; do
     for stride_idx in {0..0}; do
       echo "hostname; cd $Directory ; /home/kk/P3.10/bin/python3 cnn_training.py --idx-conv-out-channels-list $out_channels_idx --idx-conv-kernel-sizes $kernel_size_idx --idx-conv-stride-sizes $stride_idx -s \$JOB_ID" | qsub -o $Directory/argh_log_classic -j y -p $Priority -q gp4u,gp3u -N ClassicTraining

Classic_contour_net_shallow/cnn_training.py

@@ -29,7 +29,10 @@ def main(
 ) -> None:
     config_filenname = "config.json"
     with open(config_filenname, "r") as file_handle:
-        config = json.loads(jsmin(file_handle.read()))
+        file_contents = file_handle.read()
+        f_contents = jsmin(file_contents)
+        config = json.loads(f_contents)
+        # config = json.loads(jsmin(file_handle.read()))
 
     # get model information:
     output_channels = config["conv_out_channels_list"][idx_conv_out_channels_list]
@@ -81,6 +84,7 @@ def main(
         use_adam=bool(config["use_adam"]),
         use_plot_intermediate=bool(config["use_plot_intermediate"]),
         leak_relu_negative_slope=float(config["leak_relu_negative_slope"]),
+        switch_leakyR_to_relu=bool(config["switch_leakyR_to_relu"]),
         scheduler_verbose=bool(config["scheduler_verbose"]),
         scheduler_factor=float(config["scheduler_factor"]),
         precision_100_percent=int(config["precision_100_percent"]),
@@ -118,6 +122,7 @@ def run_network(
     use_adam: bool,
     use_plot_intermediate: bool,
     leak_relu_negative_slope: float,
+    switch_leakyR_to_relu: bool,
     scheduler_verbose: bool,
     scheduler_factor: float,
     precision_100_percent: int,
@@ -129,6 +134,9 @@ def run_network(
     device: torch.device = torch.device(device_str)
     torch.set_default_dtype(torch.float32)
 
+    # switch to relu if using leaky relu
+    switched_to_relu: bool = False
+
     # -------------------------------------------------------------------
     logger.info("-==- START -==-")
 
@@ -362,8 +370,19 @@ def run_network(
 
         # stop learning: done
         if round(previous_test_acc, precision_100_percent) == 100.0:
-            logger.info("100% test performance reached. Stop training.")
+            if activation_function == "leaky relu":
-            break
+                if switch_leakyR_to_relu and not switched_to_relu:
+                    logger.info(
+                        "100% test performance reached. Switching to LeakyReLU slope 0.0."
+                    )
+                    for name, module in model.named_children():
+                        if isinstance(module, torch.nn.LeakyReLU):
+                            module.negative_slope = 0.0
+                    logger.info(model)
+                    switched_to_relu = True
+            else:
+                logger.info("100% test performance reached. Stop training.")
+                break
 
     if use_plot_intermediate:
         plot_intermediate(

Classic_contour_net_shallow/config.json

@@ -19,6 +19,7 @@
     "save_ever_x_epochs": 10, // (10)
     "activation_function": "leaky relu", // tanh, relu, (leaky relu), none
     "leak_relu_negative_slope": 0.1, // (0.1)
+    "switch_leakyR_to_relu": true,
     // LR Scheduler ->
     "use_scheduler": true, // (true), false
     "scheduler_verbose": true,
@@ -34,10 +35,40 @@
     "condition": "Coignless",
     "scale_data": 255.0, // (255.0),
     "conv_out_channels_list": [
+        [
+            32,
+            8,
+            8
+        ],  
+        [
+            8,
+            8,
+            8
+        ], 
+        [
+            6,
+            8,
+            8
+        ], 
+        [
+            4,
+            8,
+            8
+        ], 
         [
             3,
             8,
             8
+        ], 
+        [
+            2,
+            8,
+            8
+        ], 
+        [
+            1,
+            8,
+            8
         ]  
     ],
     "conv_kernel_sizes": [

Classic_contour_net_shallow/functions/plot_intermediate.py

@@ -21,7 +21,9 @@ def plot_intermediate(
     assert len(train_accuracy) == len(test_losses)
 
     # legend:
-    pattern = r"(outChannels\[\d+(?:, \d+)*\]_kernelSize\[\d+(?:, \d+)*\]_)(\w+)(?=_stride)"
+    pattern = (
+        r"(outChannels\[\d+(?:, \d+)*\]_kernelSize\[\d+(?:, \d+)*\]_)([^_]+)(?=_stride)"
+    )
     matches = re.findall(pattern, save_name)
     legend_label = "".join(["".join(match) for match in matches])
 
@@ -35,8 +37,8 @@ def plot_intermediate(
     plt.figure(figsize=[12, 7])
     plt.subplot(2, 1, 1)
 
-    plt.plot(x, np.array(train_accuracy), label="Train: " + str(legend_label))
+    plt.plot(x, np.array(train_accuracy), label="Train " + str(legend_label))
-    plt.plot(x, np.array(test_accuracy), label="Test: " + str(legend_label))
+    plt.plot(x, np.array(test_accuracy), label="Test " + str(legend_label))
     plt.title("Training and Testing Accuracy", fontsize=18)
     plt.xlabel("Epoch", fontsize=18)
     plt.ylabel("Accuracy (\\%)", fontsize=18)
@@ -53,8 +55,8 @@ def plot_intermediate(
 
     # losses
     plt.subplot(2, 1, 2)
-    plt.plot(x, np.array(train_losses), label="Train: " + str(legend_label))
+    plt.plot(x, np.array(train_losses), label="Train " + str(legend_label))
-    plt.plot(x, np.array(test_losses), label="Test: " + str(legend_label))
+    plt.plot(x, np.array(test_losses), label="Test " + str(legend_label))
     plt.title("Training and Testing Losses", fontsize=18)
     plt.xlabel("Epoch", fontsize=18)
     plt.ylabel("Loss", fontsize=18)

Classic_contour_net_shallow/inspect/plot_weights.py

@@ -13,7 +13,7 @@ from functions.make_cnn import make_cnn  # noqa
 device = torch.device("cpu")
 
 # path to NN
-nn = "ArghCNN_numConvLayers3_outChannels[6, 8, 8]_kernelSize[7, 15]_leaky relu_stride1_trainFirstConvLayerTrue_seed287302_Coignless_801Epoch_2807-0857.pt"
+nn = "ArghCNN_numConvLayers3_outChannels[3, 8, 8]_kernelSize[7, 15]_leaky relu_stride1_trainFirstConvLayerTrue_seed290415_Coignless_1307Epoch_3107-0912.pt"
 PATH = f"../trained_models/{nn}"
 
 # load and evaluate model

Classic_contour_net_shallow/inspect_weights_conv_x.py

@@ -4,12 +4,13 @@ import matplotlib.pyplot as plt
 import os
 import glob
 from natsort import natsorted
+import sys
 
 # import numpy as np
 
-layer_id: int = 3
+layer_id: int = int(sys.argv[1])
-scale_each_inner: bool = False
+scale_each_inner: bool = True
-scale_each_outer: bool = False
+scale_each_outer: bool = True
 
 model_path: str = "trained_models"
 filename_list: list = natsorted(glob.glob(os.path.join(model_path, str("*.pt"))))
@@ -47,12 +48,12 @@ v_max_abs = torch.abs(weight_grid[0, ...]).max()
 plt.subplot(3, 1, (1, 2))
 plt.imshow(
     weight_grid[0, ...],
-    vmin=-v_max_abs,
+#    vmin=-v_max_abs,
-    vmax=v_max_abs,
+#    vmax=v_max_abs,
-    cmap="cool",
+    cmap="hot",
 )
 plt.axis("off")
-plt.colorbar()
+#plt.colorbar()
 plt.title("Weights")
 
 plt.subplot(3, 1, 3)

Classic_contour_net_shallow/performance_pfinkel_plots.py

@@ -132,7 +132,7 @@ if __name__ == "__main__":
     )
     assert selection_file_id < len(list_filenames)
     # model_filename: str = str(list_filenames[selection_file_id])
-    model_filename: str = "./trained_models/ArghCNN_numConvLayers3_outChannels[6, 8, 8]_kernelSize[7, 15]_leaky relu_stride1_trainFirstConvLayerTrue_seed287302_Coignless_801Epoch_2807-0857.pt"
+    model_filename: str = "./trained_models/ArghCNN_numConvLayers3_outChannels[3, 8, 8]_kernelSize[7, 15]_leaky relu_stride1_trainFirstConvLayerTrue_seed290415_Coignless_1307Epoch_3107-0912.pt"
     logger.info(f"Using model file: {model_filename}")
 
     # shorter saving name: