Brrrr

source/merge_datasets.py

@@ -2,16 +2,17 @@ import pandas as pd
 
 n = 2
 total_n = 16
+
 # list of all the csv files
 csv_files = [
-    "en-wikihow-qa-dataset-2k.csv",
-    "ru-wikihow-qa-dataset-2k.csv",
-    "pt-wikihow-qa-dataset-2k.csv",
-    "nl-wikihow-qa-dataset-2k.csv",
-    "it-wikihow-qa-dataset-2k.csv",
-    "fr-wikihow-qa-dataset-2k.csv",
-    "es-wikihow-qa-dataset-2k.csv",
-    "de-wikihow-qa-dataset-2k.csv"
+    f"en-wikihow-qa-dataset-{n}k.csv",
+    f"ru-wikihow-qa-dataset-{n}k.csv",
+    f"pt-wikihow-qa-dataset-{n}k.csv",
+    f"nl-wikihow-qa-dataset-{n}k.csv",
+    f"it-wikihow-qa-dataset-{n}k.csv",
+    f"fr-wikihow-qa-dataset-{n}k.csv",
+    f"es-wikihow-qa-dataset-{n}k.csv",
+    f"de-wikihow-qa-dataset-{n}k.csv"
 ]
 
 # create an empty dataframe to hold the merged data
@@ -22,5 +23,6 @@ for file in csv_files:
     data = pd.read_csv(file)
     merged_data = pd.concat([merged_data, data], ignore_index=True)
 
-# write the merged data to a new csv file
-merged_data.to_csv("multilingial-wikihow-qa-dataset-16k.csv", index=False)
+# write the merged data to new csv and parquet files
+merged_data.to_csv(f"multilingial-wikihow-qa-dataset-{total_n}k.csv", index=False)
+merged_data.to_parquet(f"multilingial-wikihow-qa-dataset-{total_n}k.parquet", row_group_size=100, engine="pyarrow", index=False)