document loader

app.py

@@ -1,8 +1,11 @@
 import streamlit as st
 from llmware.prompts import Prompt
 import requests
-import io
+import io, os, re
 import PyPDF2
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.embeddings import HuggingFaceInstructEmbeddings
+from langchain.vectorstores import FAISS
 
 def register_gguf_model():
 
@@ -29,48 +32,36 @@ def register_gguf_model():
     # prompter.load_model(your_model_name)
     return prompter
 
-def load_pdf_from_url(url):
-    r = requests.get(url)
-    f = io.BytesIO(r.content)
-    file = PyPDF2.PdfReader(f)
-    return file
-
-def load_pdf_content(pdf):
-    content = ""
-    for page in pdf.pages:
-        content += page.extract_text()
-    return content
-
-
 def main():
     st.title("BetterZila RAG Enabled LLM")
     with st.spinner("Registering Models for use..."):
         prompter = register_gguf_model()
-      
+    
+    data_path = "data/"
+    
+    # keep the select box to llama as default but give a button right below it that says select model after which the model will be loaded
+    st.sidebar.subheader("Select Model")
     model_name = st.sidebar.selectbox("Select Model", ["llama", "open_gpt4", "phi2", "mistral"])
-    with st.spinner("Loading model..."):
-        print(model_name)
+    with st.spinner("Loading Model..."):
         prompter.load_model(model_name)
-
-    with st.spinner("Loading PDF content from the assignment URL..."):
-        url = "https://pgcag.files.wordpress.com/2010/01/48lawsofpower.pdf"
-        pdf = load_pdf_from_url(url)
-        content = load_pdf_content(pdf)
-        print("Loaded PDF content")
-    st.success("PDF content loaded!")
+    st.success("Model Loaded!")
     
     queries = ['Can you give me an example from history where the enemy was crushed totally from the book?', "What's the point of making myself less accessible?", "Can you tell me the story of Queen Elizabeth I from this 48 laws of power book?"]
     
     for query in queries:
         st.subheader(f"Query: {query}")
         with st.spinner("Generating response..."):
-            print("Query: ", query)
-            resp = prompter.prompt_main(query, context=content)
-            response = resp['llm_response']
-            print("Time taken: ", response['usage']['processing_time'])
-            print("Response: ", response)
+            for file in os.listdir(data_path):
+                if file.endswith(".pdf"):
+                    source = prompter.add_source_document(data_path, file, query=None)
+                    responses = prompter.prompt_with_source(query, prompt_name="just_the_facts", temperature=0.3)
+                    for r, response in enumerate(responses):
+                        print(query, ":", re.sub("[\n]"," ", response["llm_response"]).strip())
+                    prompter.clear_source_materials()
+                    st.write(query)
+                    st.write(re.sub("[\n]"," ", response["llm_response"]).strip())
         st.success("Response generated!")
-        st.write(response)
+        
         
 if __name__ == "__main__":
     main()

requirements.txt

@@ -1,4 +1,5 @@
 llmware
 streamlit
 requests
-PyPDF2
+PyPDF2
+langchain