testing with llama3

src/app.py

@@ -2,7 +2,7 @@ from interface import create_demo
 from pdfchatbot import PDFChatBot
 
 # Create Gradio interface
-demo, chat_history, show_img, txt, submit_button, uploaded_pdf, slider1 = create_demo()
+demo, chat_history, show_img, txt, submit_button, uploaded_pdf, _ = create_demo()
 
 # Create PDFChatBot instance
 pdf_chatbot = PDFChatBot()
@@ -12,7 +12,7 @@ pdf_chatbot.load_tokenizer()
 # Set up event handlers
 with demo:
     # Event handler for uploading a PDF
-    uploaded_pdf.upload(pdf_chatbot.render_file, inputs=[uploaded_pdf,slider1], outputs=[show_img])
+    uploaded_pdf.upload(pdf_chatbot.render_file, inputs=[uploaded_pdf], outputs=[show_img])
 
     # Event handler for submitting text and generating response
     submit_button.click(pdf_chatbot.add_text, inputs=[chat_history, txt], outputs=[chat_history], queue=False).\

src/interface.py

@@ -11,7 +11,7 @@ def create_demo():
         with gr.Row():
             # Add sliders here
             with gr.Column():  # Adjust scale as needed
-                slider1 = gr.Slider(minimum=256, maximum=1024, value=50, label="Chunk Size")
+                slider1 = gr.Slider(minimum=256, maximum=1024, value=256, label="Chunk Size")
         with gr.Row():
             with gr.Column(scale=0.60):
                 text_input = gr.Textbox(

src/pdfchatbot.py

@@ -95,7 +95,7 @@ class PDFChatBot:
         """
         Load the vector database from the documents and embeddings.
         """
-        text_splitter = CharacterTextSplitter(chunk_size=self.chunk_size, chunk_overlap=256)
+        text_splitter = CharacterTextSplitter(chunk_size=self.chunk_size, chunk_overlap=100)
         docs = text_splitter.split_documents(self.documents)
         self.vectordb = Chroma.from_documents(docs, self.embeddings)
 
@@ -178,13 +178,11 @@ class PDFChatBot:
 
         result = self.chain({"question": query, 'chat_history': self.chat_history}, return_only_outputs=True)
         self.chat_history.append((query, result["answer"]))
-        self.page = list(result['source_documents'][0])[1][1]['page']
-
         for char in result['answer']:
             history[-1][-1] += char
         return history, " "
 
-    def render_file(self, file,chunk_size):
+    def render_file(self, file,chunk_size=256):
         """
         Renders a specific page of a PDF file as an image.