sanitize

SilverSulfide · web-flow · commit 9151e5a1ef29 · 2025-08-20T09:20:51.000+03:00
diff --git a/mock_data.py b/mock_data.py
@@ -64,7 +64,7 @@ def load_indexed_dset(path_to_bin):
     mock_docs = generate_mock_documents(num_docs, max_doc_len)
 
     # instruction tokens
-    tokens_instruct = [111111, 1337, 420, 1337, 1337, 111111]
+    tokens_instruct = [111111, 222, 333, 444, 333, 222, 111111]
 
     # iterate through the documents
     for temp_tokens in tqdm(mock_docs):
