Change by RAPIDS instead

app.py

@@ -6,8 +6,10 @@ from torch import cuda
 from gradio_huggingfacehub_search import HuggingfaceHubSearch
 from bertopic import BERTopic
 from bertopic.representation import KeyBERTInspired
-from umap import UMAP
-from hdbscan import HDBSCAN
+
+from cuml.manifold import UMAP
+from cuml.cluster import HDBSCAN
+
 from sklearn.feature_extraction.text import CountVectorizer
 
 from sentence_transformers import SentenceTransformer
@@ -32,8 +34,8 @@ logging.basicConfig(
     level=logging.INFO, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
 )
 
-MAX_ROWS = 5_000
-CHUNK_SIZE = 1_000
+MAX_ROWS = 10_000
+CHUNK_SIZE = 2_000
 
 
 session = requests.Session()