Update app.py

app.py

@@ -29,7 +29,8 @@ def load_doc(list_file_path, chunk_size, chunk_overlap):
     return text_splitter.split_documents(pages)
 
 def create_db(splits, collection_name):
-    embedding = HuggingFaceEmbeddings()
+    # Use the lightweight MiniLM model for embeddings
+    embedding = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
     new_client = chromadb.EphemeralClient()
     return Chroma.from_documents(documents=splits, embedding=embedding, client=new_client, collection_name=collection_name)
 
@@ -112,10 +113,10 @@ def demo():
             document = gr.Files(height=100, file_count="multiple", file_types=["pdf"], interactive=True, label="Carica i tuoi documenti PDF")
         
         with gr.Tab("Passo 2 - Elabora Documenti"):
-            db_btn = gr.Radio(["ChromaDB"], label="Tipo di database vettoriale", value="ChromaDB", type="index")
+            db_btn = gr.Radio(["ChromaDB (MiniLM Embedding)"], label="Tipo di database vettoriale", value="ChromaDB (MiniLM Embedding)", type="index")
             with gr.Accordion("Opzioni Avanzate - Divisione del testo del documento", open=False):
-                slider_chunk_size = gr.Slider(100, 1000, 1000, step=20, label="Dimensione del chunk")
-                slider_chunk_overlap = gr.Slider(10, 200, 100, step=10, label="Sovrapposizione del chunk")
+                slider_chunk_size = gr.Slider(100, 1000, 500, step=20, label="Dimensione del chunk")
+                slider_chunk_overlap = gr.Slider(10, 200, 50, step=10, label="Sovrapposizione del chunk")
             db_progress = gr.Textbox(label="Inizializzazione del database vettoriale", value="Nessuna")
             db_btn = gr.Button("Genera database vettoriale")