Use 41m embeddings

app.py

@@ -9,11 +9,11 @@ import faiss
 from usearch.index import Index
 
 # Load titles and texts
-title_text_dataset = load_dataset("mixedbread-ai/wikipedia-2023-11-embed-en-pre-1", split="train").select_columns(["title", "text"])
+title_text_dataset = load_dataset("mixedbread-ai/wikipedia-embed-en-2023-11", split="train").select_columns(["title", "text"])
 
 # Load the int8 and binary indices. Int8 is loaded as a view to save memory, as we never actually perform search with it.
-int8_view = Index.restore("wikipedia_int8_usearch_1m.index", view=True)
-binary_index: faiss.IndexBinaryFlat = faiss.read_index_binary("wikipedia_ubinary_faiss_1m.index")
+int8_view = Index.restore("wikipedia_int8_usearch_50m.index", view=True)
+binary_index: faiss.IndexBinaryFlat = faiss.read_index_binary("wikipedia_ubinary_faiss_50m.index")
 
 # Load the SentenceTransformer model for embedding the queries
 model = SentenceTransformer(