reduce memory usage.

src/wikiq/__init__.py

@@ -493,9 +493,9 @@ class WikiqParser:
                     num_context_lines=1000000,
                     max_word_level_diff_complexity=-1,
                     moved_paragraph_detection_cutoff=-1,
-                    words_cache_capacity=5000,
-                    diff_cache_capacity=5000,
-                    stats_cache_capacity=5000,
+                    words_cache_capacity=2000,
+                    diff_cache_capacity=2000,
+                    stats_cache_capacity=2000,
                 )
 
             while not on_last_batch:
@@ -875,7 +875,7 @@ def main():
     parser.add_argument(
         "--batch-size",
         dest="batch_size",
-        default=16000,
+        default=8000,
         type=int,
         help="How many revisions to process in each batch. This ends up being the Parquet row group size",
     )