Add files via upload

get_perf.py

@@ -1,47 +1,47 @@
 import os
-
 os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
-import matplotlib.pyplot as plt
+
+which_scalar = "Test Error"
 
 from tensorboard.backend.event_processing import event_accumulator
 import numpy as np
-import json
-from jsmin import jsmin
 import glob
 
-# -------------------------------
+log_paths: str = "Log*"
+log_paths_list = glob.glob(log_paths)
+assert len(log_paths_list) > 0
 
-filename:str = "def.json"
+for path in log_paths_list:
-with open(filename) as json_file:
+    print(path)
-    minified = jsmin(json_file.read())
+    temp = path.split("_")
-data = json.loads(minified)
+    if len(temp) == 2:
+        parameter:str | None = temp[-1]
+    else: 
+        parameter = None
 
+    # ----------------------
+    temp = glob.glob(path)
+    assert len(temp) == 1
 
+    acc = event_accumulator.EventAccumulator(path)
+    acc.Reload()
 
-# -------------------------------
+    # Check if the requested scalar exists 
+    available_scalar = acc.Tags()["scalars"]
+    # available_histograms = acc.Tags()["histograms"]
+    available_scalar.index(which_scalar)
 
+    te = acc.Scalars(which_scalar)
 
-path_runs: str = "./Log/*"  
+    np_temp = np.zeros((len(te), 2))
 
-temp = glob.glob(path_runs)
+    for id in range(0, len(te)):
-assert len(temp) == 1
+        np_temp[id, 0] = te[id][1]
-path = temp[0]
+        np_temp[id, 1] = te[id][2]
+    print(np_temp)
 
-
+    if parameter is not None:
-acc = event_accumulator.EventAccumulator(path)
+        np.save(f"result_{parameter}.npy", np_temp)
-acc.Reload()
+    else:
-
+        np.save(f"result.npy", np_temp)
-available_scalar = acc.Tags()["scalars"]
-available_histograms = acc.Tags()["histograms"]
-
-which_scalar = "Test Error"
-te = acc.Scalars(which_scalar)
-
-temp = []
-for te_item in te:
-    temp.append((te_item[1], te_item[2]))
-temp = np.array(temp)
-
-print(temp)
-np.save(f"test_error.npy", temp)
 

train_it.py

@@ -30,8 +30,10 @@ from network.loop_train_test import (
     run_lr_scheduler,
     loop_test_reconstruction,
 )
+
 from network.SbSReconstruction import SbSReconstruction
 from network.InputSpikeImage import InputSpikeImage
+from network.SbSLayer import SbSLayer
 
 from torch.utils.tensorboard import SummaryWriter
 
@@ -155,6 +157,8 @@ if order_id is not None:
     if isinstance(network[0], InputSpikeImage) is True:
         network[0].number_of_spikes = number_of_spikes_in_input_layer
 
+    if isinstance(network[0], SbSLayer) is True:
+        network[0]._number_of_spikes = number_of_spikes_in_input_layer
 
 last_test_performance: float = -1.0
 with torch.no_grad():