Fixed loading script

src/loading.py

 import tensorflow as tf
 
-
 def load_dataset(filename, label):
     # Loading the dataset file
-    dataset = tf.data.experimental.make_csv_dataset(filename, 200,
-     select_columns=[2],column_defaults =  [ tf.float32,  ],num_epochs=100000, 
-     header=True, field_delim=" ")
+    dataset = tf.data.experimental.CsvDataset(filename,   [tf.float32, ], select_cols=[2], header=True, field_delim=" ")
+
+    # Reshape the incoming data
+    dataset = dataset.batch(200)
 
     # Create labelled vector
-    dataset = dataset.map(lambda x: (x.values(),label))
+    dataset = dataset.map(lambda x: (tf.reshape(x,(200,1)),label))
 
     return dataset
 

src/main.py

@@ -29,14 +29,14 @@ def main(argv):
         in_layer = (Nbeads,1)
 
     datasets = []
-    for knot in knots:
-        datasets.append(load_dataset("/storage/cmstore02/groups/TAPLab/knots_database/" + str(knot) + "/N200/lp10/SIGWRITHE/3DSignedWrithe_" + knot + ".dat.lp10.dat.nos", knot))
+    for i, knot in enumerate(knots):
+        datasets.append(load_dataset("/storage/cmstore02/groups/TAPLab/knots_database/" + str(knot) + "/N200/lp10/SIGWRITHE/3DSignedWrithe_" + knot + ".dat.lp10.dat.nos", i))
 
-    dataset = combine_datasets(datasets)
+    dataset = tf.data.experimental.sample_from_datasets(datasets)
 
-    for element in dataset.as_numpy_iterator():
-        print(element)
-        break
+    dataset = dataset.shuffle(buffer_size=1)
+
+    dataset = dataset.repeat()
 
     if net == "FFNN":
         model = setup_NN(
@@ -59,7 +59,7 @@ def main(argv):
 
     # Early Stopping Callback
     es = tf.keras.callbacks.EarlyStopping(
-        monitor="val_loss",
+        monitor="loss",
         mode="min",
         verbose=1,
         patience=10,
@@ -84,20 +84,25 @@ def main(argv):
     mc = tf.keras.callbacks.ModelCheckpoint(
         filepath=checkpoint_filepath,
         save_weights_only=False,
-        monitor="val_loss",
+        monitor="loss",
         mode="min",
         save_best_only=True,
     )
 
     cb_list = [es, mc]
     bs = 256
+    dataset = dataset.batch(bs)
+
+    for i,element in enumerate(dataset.as_numpy_iterator()):
+        print(element)
+        if i>=1:
+            break
 
     model.fit(
         dataset,
-        epochs=1,
-        batch_size=bs,
-        verbose=0,
-        # validation_split=1 / 5,
+        steps_per_epoch=300,
+        epochs=20,
+        verbose=1,
         callbacks=cb_list,
     )