Update app.py

app.py

@@ -93,20 +93,31 @@ st.markdown("""
 """, unsafe_allow_html=True)
 
 # ----------------- لود PDF و ساخت ایندکس -----------------
+class TogetherEmbeddings(Embeddings):
+    def __init__(self, model_name: str, api_key: str):
+        self.model_name = model_name
+        self.client = Together(api_key=api_key)
+
+    def embed_documents(self, texts: List[str]) -> List[List[float]]:
+        response = self.client.embeddings.create(model=self.model_name, input=texts)
+        return [item.embedding for item in response.data]
+
+    def embed_query(self, text: str) -> List[float]:
+        return self.embed_documents([text])[0]
+
 @st.cache_resource
 def get_pdf_index():
     with st.spinner('📄 در حال پردازش فایل PDF...'):
-        pdf_loader = PyPDFLoader('test1.pdf')
-
+        loader = [PyPDFLoader('test1.pdf')]
         embeddings = TogetherEmbeddings(
+            model_name="togethercomputer/m2-bert-80M-8k-retrieval",
             api_key="0291f33aee03412a47fa5d8e562e515182dcc5d9aac5a7fb5eefdd1759005979"
         )
+        return VectorstoreIndexCreator(
+            embedding=embeddings,
+            text_splitter=RecursiveCharacterTextSplitter(chunk_size=300, chunk_overlap=0)
+        ).from_loaders(loader)
 
-        index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([pdf_loader])
-        
-        return index
-
-# ----------------- بارگذاری دیتا -----------------
 index = get_pdf_index()
 
 llm = ChatOpenAI(
@@ -152,7 +163,6 @@ if st.session_state.pending_prompt:
         thinking.empty()
         full_response = ""
         placeholder = st.empty()
-
         for word in answer.split():
             full_response += word + " "
             placeholder.markdown(full_response + "▌")