Merge pull request #5 from ptptiwari/main

More improvements in the chatbot code and few minor fixes

python-rag-chatbot/README.md

@@ -10,11 +10,16 @@
 ## TODO instructions
 
 - setup ~/.oci/config
-- set yourcompartmentid
+- oci session authenticate and make sure authentication is successful
+- use region where the OCI gen AI is available. Currently we are using chicago region. 
+- run pip install -r requirements.txt
+- check you config_rag.py file and make sure your api endpoint belong to chicago region and db which you want to use like chroma db or oracle db
+- set your compartment_id ocid inside the file i.e. init_rag_streamlit_exp.py and init_rag.py file
+- Changing the db type you need to modify at config file and you see the logic inside create_vector_store
 - podman run -d --name 23ai -p 1521:1521 -e ORACLE_PWD=<password> -v oracle-volume:/Users/pparkins/oradata container-registry.oracle.com/database/free:latest
 - create/config vector tablespace and user
 - add oracle database info for use in init_rag_streamlit.py / init_rag_streamlit_exp.py
-- run run_oracle_bot.sh /run_oracle_bot_exp.sh
+- run ./run_oracle_bot_exp.sh
 
 
 ## Documentation

python-rag-chatbot/config_rag.py

@@ -4,18 +4,14 @@
 DEBUG = False
 
 # book to use for augmentation
-# BOOK1 = "APISpec.pdf"
+
 BOOK1 = "pdfFiles/sharding-adg-addshard-cookbook-3610618.pdf"
 BOOK2 = "pdfFiles/globally-distributed-autonomous-database.pdf"
-# BOOK4 = "OnBoardingGuide.pdf"
-# BOOK5 = "CreateWorkFlow.pdf"
-# BOOK6 = "Team Onboarding.pdf"
-# BOOK7 = "workflow.pdf"
 BOOK3 = "pdfFiles/oracle-database-23c.pdf"
 BOOK4 = "pdfFiles/oracle-globally-distributed-database-guide.pdf"
 BOOK5 = "pdfFiles/Oracle True cache.pdf"
 BOOK6 = "pdfFiles/Trobleshooting.pdf"
-# BOOK12 = "OsdCode.pdf"
+
 
 BOOK_LIST = [BOOK1, BOOK2, BOOK3, BOOK4, BOOK5, BOOK6]
 

python-rag-chatbot/init_rag.py

@@ -49,7 +49,7 @@
 
 # private configs
 CONFIG_PROFILE = "DEFAULT"
-COMPARTMENT_OCID = "ocid1.compartment.oc1..yourcompartmentid"
+COMPARTMENT_OCID = "ocid1.compartment.yourcomparmentid"
 oci_config = oci.config.from_file("~/.oci/config", CONFIG_PROFILE)
 COHERE_API_KEY = oci_config['security_token_file']
 DEBUG = True
@@ -155,7 +155,7 @@ def initialize_rag_chain():
 
         llm = OCIGenAI(
             service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
-            compartment_id="ocid1.compartment.oc1..yourcompartmentid",
+            compartment_id="ocid1.compartment.oc1..yourcompartment",
             model_kwargs={"max_tokens": 1000},
             auth_type='SECURITY_TOKEN',
         )

python-rag-chatbot/init_rag_streamlit.py

@@ -62,7 +62,7 @@
 
 # private configs
 CONFIG_PROFILE = "DEFAULT"
-COMPARTMENT_OCID = "ocid1.compartment.oc1..aaaaaaaajdyhd7dqnix2avhlckbhhkkcl3cujzyuz6jzyzonadca3i66pqjq"
+COMPARTMENT_OCID = "ocid1.compartment.oc1..yourcompartment"
 oci_config = oci.config.from_file("~/.oci/config", CONFIG_PROFILE)
 COHERE_API_KEY = oci_config['key_file']
 #

python-rag-chatbot/init_rag_streamlit_exp.py

@@ -63,7 +63,7 @@
 
 # private configs
 CONFIG_PROFILE = "DEFAULT"
-COMPARTMENT_OCID = "ocid1.compartment.oc1..yourcompartmentid"
+COMPARTMENT_OCID = "ocid1.compartment.oc1..yourcompartment"
 oci_config = oci.config.from_file("~/.oci/config", CONFIG_PROFILE)
 COHERE_API_KEY = oci_config['security_token_file']
 
@@ -134,6 +134,7 @@ def split_in_chunks(all_pages):
     splits = post_process(splits)
 
     print(f"Splitted the pdf in {len(splits)} chunks...")
+    print("Size of splits: " + str(text_splitter.__sizeof__()) + "bytes")
 
     return splits
 
@@ -252,7 +253,7 @@ def build_llm(llm_type):
         llm = OCIGenAI(
             model_id="cohere.command",
             service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
-            compartment_id="ocid1.compartment.oc1..yourcompartmentid",
+            compartment_id="ocid1.compartment.oc1..yourCompartment",
             model_kwargs={"max_tokens": 200},
             auth_type='SECURITY_TOKEN',
         )
@@ -284,13 +285,14 @@ def initialize_rag_chain():
 
     # 3. Load embeddings model
     embedder = create_cached_embedder()
-
+    print("Size of splits---: " + str(document_splits.__sizeof__()) + "bytes")
     # 4. Create a Vectore Store and store embeddings
     vectorstore = create_vector_store(VECTOR_STORE_NAME, document_splits, embedder)
 
     # 5. Create a retriever
     # increased num. of docs to 5 (default to 4)
     # added optionally a reranker
+
     retriever = create_retriever(vectorstore)
 
     # 6. Build the LLM

python-rag-chatbot/oracle_bot_exp.py

@@ -1,16 +1,18 @@
 import streamlit as st
-import time
 import traceback
 import sys
-
 from init_rag_streamlit_exp import initialize_rag_chain, get_answer
-
 from streamlit_feedback import streamlit_feedback
 
 def process_feedback(feedback_value):
-    st.write("Feedback value:", feedback_value)
-    with open("feedback.txt", "a", encoding="utf-8") as f:
-        f.write(f"{feedback_value}\n")
+    st.write("Processing feedback value:", feedback_value)  # Debugging output
+    try:
+        with open("feedback.txt", "a", encoding="utf-8") as f:
+            f.write(f"{feedback_value}\n")
+        st.write("Feedback successfully written to file.")  # Debugging output
+    except Exception as e:
+        st.error(f"Error writing to file: {e}")
+        traceback.print_exc()
 
 def reset_conversation():
     st.session_state.messages = []
@@ -53,6 +55,7 @@ def reset_conversation():
                 if not st.session_state.feedback_rendered:
                     def _submit_feedback(feedback_value, *args, **kwargs):
                         print("Feedback submitted:", feedback_value, file=sys.stderr)  # Redirect to stderr
+                        st.write("Feedback value received for submission:", feedback_value)  # Debugging output
                         process_feedback(feedback_value)
                         st.session_state.feedback_rendered = False