diff --git a/prosit_RT_ori.py b/prosit_RT_ori.py
new file mode 100644
index 0000000000000000000000000000000000000000..b07b71d29e5b834d0ddb6b5f10065c05ca9fc254
--- /dev/null
+++ b/prosit_RT_ori.py
@@ -0,0 +1,58 @@
+import os
+import wandb as wdb
+import numpy as np
+import pandas as pd
+import dlomix
+from dlomix import constants, data, eval, layers, models, pipelines, reports, utils
+from dlomix.models import RetentionTimePredictor
+from dlomix.data import RetentionTimeDataset
+from dlomix.eval import TimeDeltaMetric
+from dlomix.reports import RetentionTimeReport
+
+batch_size = 1024
+os.environ["WANDB_API_KEY"] = 'b4a27ac6b6145e1a5d0ee7f9e2e8c20bd101dccd'
+os.environ["WANDB_MODE"] = "offline"
+
+config = {
+    "model": "RT prediction GRU/selfAtt+ GRU",
+    "batch_size":batch_size,
+}
+
+wdb.init(project="RT prediction")
+
+TRAIN_DATAPATH = '/database/data.csv'
+
+data = pd.read_csv('/database/data.csv')
+data_train = data[data.state == 'train' or data.state == 'validation']
+data_holdout = data[data.state == 'holdout']
+
+data_train.to_csv('/database/data_train.csv')
+data_holdout.to_csv('/database/data_holdout.csv')
+
+rtdata = RetentionTimeDataset(data_source='/database/data_train.csv',
+                              seq_length=30, batch_size=batch_size, val_ratio=0.2, test=False)
+
+model = RetentionTimePredictor(seq_length=30)
+
+model.compile(optimizer='adam',
+              loss='mse',
+              metrics=['mean_absolute_error', TimeDeltaMetric()])
+
+history = model.fit(rtdata.train_data,
+                    validation_data=rtdata.val_data,
+                    epochs=20)
+
+test_rtdata = RetentionTimeDataset(data_source='/database/data_holdout.csv',
+                              seq_length=30, batch_size=1024, test=True)
+
+predictions = model.predict(test_rtdata.test_data)
+
+# we use ravel from numpy to flatten the array (since it comes out as an array of arrays)
+predictions = predictions.ravel()
+
+test_targets = test_rtdata.get_split_targets(split="test")
+
+report = RetentionTimeReport(output_path="./output", history=history)
+report.calculate_r2(test_targets, predictions)
+
+wdb.finish()
\ No newline at end of file