Shuffle the data

the_pile_github.py

@@ -182,9 +182,10 @@ class SmartContracts(datasets.GeneratorBasedBuilder):
         #data = pd.read_parquet(filepath)
         key = 0
         for path in files:
-            data = pd.read_parquet(path)
+            df = pd.read_parquet(path)
+            df = df.sample(frac = 1, random_state = 42)
 
-            for row in data.itertuples():
+            for row in df.itertuples():
 
                 if split_key == "train":
                     yield key, {