fixed stuff

app.py

@@ -10,8 +10,10 @@ import numpy as np
 
 embedding_client = InferenceClient(model="sentence-transformers/all-MiniLM-L6-v2")
 
-def embed(texts):
-    return embedding_client.feature_extraction(texts)
+def embed_texts(texts):
+    if isinstance(texts, str):
+        texts = [texts]
+    return np.array(embedding_client.feature_extraction(texts))
 
 
 
@@ -22,8 +24,6 @@ def embed(texts):
 with open("gita.txt", "r", encoding="utf-8") as f:
     raw_text = f.read()
 
-doc_embeddings = np.array(embed(documents))
-
 def chunk_text(text, chunk_size=500, overlap=50):
     chunks = []
     start = 0
@@ -33,6 +33,9 @@ def chunk_text(text, chunk_size=500, overlap=50):
         start += chunk_size - overlap
     return chunks
 
+documents = chunk_text(raw_text)
+doc_embeddings = embed_texts(documents)
+
 # Embedding model (small + free)
 #embedder = SentenceTransformer("all-MiniLM-L6-v2")
 #doc_embeddings = #embedder.encode(documents)
@@ -40,7 +43,7 @@ def chunk_text(text, chunk_size=500, overlap=50):
 #doc_embeddings = embedder.encode(documents)
 
 def retrieve(query, top_k=4):
-    query_embedding = embedder.encode([query])[0]
+    query_embedding = embed_texts(query)[0]
     scores = np.dot(doc_embeddings, query_embedding)
     top_indices = np.argsort(scores)[-top_k:][::-1]
     results = [documents[i] for i in top_indices]