Update app.py

app.py

@@ -43,14 +43,10 @@ HF_TOKEN = os.environ.get("HF_TOKEN")
 if not HF_TOKEN:
     raise ValueError("HF_TOKEN not found in environment variables")
 
-# Custom UI
-TITLE = "<h1><center>Multi-Document RAG with LLama 3.1-8B Model</center></h1>"
-DESCRIPTION = """
-<center>
-<p>Upload PDF or text files to get started!</p>
-<p>After asking question wait for RAG system to get relevant nodes and pass to LLM</p>
-</center>
-"""
+# --- UI Settings ---
+TITLE = "<h1 style='text-align:center; margin-bottom: 20px;'>Local RAG Llama 3.1 8B</h1>"
+# Description 제거 (사용자 요청)
+
 CSS = """
 .upload-section {
     max-width: 400px;
@@ -226,8 +222,8 @@ def create_or_update_index(files, request: gr.Request):
         }
     
     node_parser = HierarchicalNodeParser.from_defaults(
-        chunk_sizes=[2048, 512, 128],  
-        chunk_overlap=20         
+        chunk_sizes=[2048, 512, 128],
+        chunk_overlap=20
     )
     logger.info(f"Parsing {len(new_documents)} documents into hierarchical nodes")
     new_nodes = node_parser.get_nodes_from_documents(new_documents)
@@ -407,7 +403,6 @@ def stream_chat(
             partial_response += new_text
             updated_history[-1]["content"] = partial_response
             yield updated_history
-        output_ids = global_tokenizer.encode(partial_response, return_tensors="pt")
         yield updated_history
     except GeneratorExit:
         stop_event.set()
@@ -416,14 +411,14 @@ def stream_chat(
 
 def create_demo():
     with gr.Blocks(css=CSS, theme=gr.themes.Soft()) as demo:
+        # Title only
         gr.HTML(TITLE)
-        gr.HTML(DESCRIPTION)
         
         with gr.Row(elem_classes="main-container"):
             with gr.Column(elem_classes="upload-section"):
                 file_upload = gr.File(
                     file_count="multiple",
-                    label="Drag and Drop Files Here",
+                    label="Drag & Drop PDF/TXT Files Here",
                     file_types=[".pdf", ".txt"],
                     elem_id="file-upload"
                 )
@@ -446,7 +441,7 @@ def create_demo():
             with gr.Column(elem_classes="chatbot-container"):
                 chatbot = gr.Chatbot(
                     height=500,
-                    placeholder="Chat with your documents here... Type your question below.",
+                    placeholder="Chat with your documents...",
                     show_label=False,
                     type="messages"
                 )
@@ -462,7 +457,7 @@ def create_demo():
                 
                 with gr.Accordion("Advanced Settings", open=False):
                     system_prompt = gr.Textbox(
-                        value="As a knowledgeable assistant, your task is to provide detailed and context-rich answers based on the relevant information from all uploaded documents. When information is sourced from multiple documents, summarize the key points from each and explain how they relate, noting any connections or contradictions. Your response should be thorough, informative, and easy to understand.",
+                        value="As a knowledgeable assistant, provide detailed answers using the relevant information from all uploaded documents.",
                         label="System Prompt",
                         lines=3
                     )
@@ -472,7 +467,7 @@ def create_demo():
                             minimum=0,
                             maximum=1,
                             step=0.1,
-                            value=0.9,  
+                            value=0.9,
                             label="Temperature"
                         )
                         max_new_tokens = gr.Slider(
@@ -553,10 +548,9 @@ def create_demo():
                     ],
                     outputs=chatbot
                 )
-
     return demo
 
 if __name__ == "__main__":
     initialize_model_and_tokenizer()
     demo = create_demo()
-    demo.launch()
+    demo.launch()