Replace .csv to .parquet

README.md

@@ -12,17 +12,17 @@ configs:
 - config_name: default
   data_files:
   - split: flat_5b
-    path: "flat/5b/*.csv"
+    path: "flat/5b/*.parquet"
   - split: flat_500m
-    path: "flat/500m/*.csv"
+    path: "flat/500m/*.parquet"
   - split: flat_50m
-    path: "flat/50m/*.csv"
+    path: "flat/50m/*.parquet"
   - split: sequential_5b
-    path: "sequential/5b/*.csv"
+    path: "sequential/5b/*.parquet"
   - split: sequential_500m
-    path: "sequential/500m/*.csv"
+    path: "sequential/500m/*.parquet"
   - split: sequential_50m
-    path: "sequential/50m/*.csv"
+    path: "sequential/50m/*.parquet"
 ---
 
 # Yambda-5B — A Large-Scale Multi-modal Dataset for Ranking And Retrieval