Update app.py

app.py

@@ -1,121 +1,127 @@
-import streamlit as st
-from langchain import OpenAI, PromptTemplate, LLMChain
-from langchain.text_splitter import CharacterTextSplitter
-from langchain.chains.mapreduce import MapReduceChain
-from langchain.prompts import PromptTemplate
-from langchain.chat_models import AzureChatOpenAI
-from langchain.chains.summarize import load_summarize_chain
-from langchain.chains import AnalyzeDocumentChain
-from PyPDF2 import PdfReader
+%%writefile app.py
+import os
+from langchain.chains import RetrievalQA
+from langchain.llms import AzureOpenAI
 from langchain.document_loaders import TextLoader
-from langchain.indexes import VectorstoreIndexCreator
 from langchain.document_loaders import PyPDFLoader
-import os
-import openai
+from langchain.indexes import VectorstoreIndexCreator
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.embeddings import OpenAIEmbeddings
+from langchain.vectorstores import Chroma
+from langchain.chains.question_answering import load_qa_chain
+from langchain.llms import AzureOpenAI
+from langchain.chains.question_answering import load_qa_chain
+import streamlit as st
+from PIL import Image
 
+#image = Image.open('https://www.wipro.com/content/dam/nexus/en/brand/images/secondary-logo-400x276.png')
+#st.image(image, caption='Wipro impact logo')
 
-import os
+st.title("Wipro Impact | CSRD | Advisor")
 
+st.header("Enable clients to prepare for CSRD.. ")
 
-os.environ["OPENAI_API_TYPE"] = "azure"
-os.environ["OPENAI_API_VERSION"] = "2023-03-15-preview"
 
-openai.api_type = "azure"
-openai.api_base = "https://embeddinguseopenai.openai.azure.com/"
-openai.api_version = "2023-03-15-preview"
-openai.api_key = os.environ["OPENAI_API_KEY"]
- 
+genre = st.radio(
+    "Choose a CSRD focus area for AI question answering",
+    ('E1-Climate Change', 'E4-Biodiversity and ecosystem', 'S1-Own Workforce'), index=0)
 
+url = "https://www.efrag.org/Assets/Download?assetUrl=%2Fsites%2Fwebpublishing%2FSiteAssets%2F13%2520Draft%2520ESRS%2520S1%2520Own%2520workforce%2520November%25202022.pdf"
 
+if genre == 'E1-Climate Change':
+  url = "https://www.efrag.org/Assets/Download?assetUrl=%2Fsites%2Fwebpublishing%2FSiteAssets%2F08%2520Draft%2520ESRS%2520E1%2520Climate%2520Change%2520November%25202022.pdf"
+  samplequestions = ["What are the climate related indicators?", "List all the disclosure requirments listed in page 3?",  "Policies related to climate change mitigation and adaptation?", 
+  "What should the company disclose regarding Actions and resources in relation to climate change policies?" , "How should the  GHG emission reduction targets be reported?" , 
+  "Total energy consumption from non-renewable sources for high climate impact sectors should be disaggregated as ?", 
+  "Total energy consumption from renewable sources should be disaggregated as ?" , 
+  "What should be disclosed on GHG removals and GHG mitigation projects financed through carbon credits ?" , 
+  "Is it wise to invest on carbon credits? ",
+  "What is Climate change adaptation? ", "What are Decarbonisation levers?"   ]
 
+if genre == 'E4-Biodiversity and ecosystem':
+  url = "https://www.efrag.org/Assets/Download?assetUrl=%2Fsites%2Fwebpublishing%2FSiteAssets%2F11%2520Draft%2520ESRS%2520E4%2520Biodiversity%2520and%2520ecosystems%2520November%25202022.pdf"
+  samplequestions = ["What are the Biodiversity related indicators?", "List all the disclosure requirments listed in page 3?"]
+if genre == 'S1-Own Workforce':
+  url = "https://www.efrag.org/Assets/Download?assetUrl=%2Fsites%2Fwebpublishing%2FSiteAssets%2F13%2520Draft%2520ESRS%2520S1%2520Own%2520workforce%2520November%25202022.pdf" 
+  samplequestions = ["What are the social related indicators?", "List all the disclosure requirments listed in page 3?"]
 
-st.title("Wipro demo with azure cognitive 2 ")
 
 
-atemprature = st.slider('Fact vs Creative?', 0, 10, 1)
-atemprature = atemprature / 10.0
 
+sampleselectedquestion = st.selectbox(
+    'Just ask your question or start with a one of these example questions...',  samplequestions    )
 
-yourquestion = st.text_input('Your Question', 'First identify the indicators required as per EFRAG Environmental document. List these indicators. For each of these indicators, find out how Wipro is performing.')
-st.write('Your input is ', yourquestion)
+st.write(" :green[ Ask any thing on your mind...just type your question here...]")
+yourquestion = st.text_input('Your question', sampleselectedquestion)
+st.write('Your typed .. ', yourquestion)
 
 
 
 
+os.environ['OPENAI_API_TYPE'] = 'azure'
+os.environ['OPENAI_API_VERSION'] = '2023-03-15-preview'
 
-if st.button("Ask Questions "):
-    template = """
-    You are an AI assistant. 
-    {concept}
-    """
+llmgpt3 = AzureOpenAI(      deployment_name="testdavanci", model_name="text-davinci-003" )
+chain = load_qa_chain(llm=llmgpt3, chain_type="map_reduce")
 
-    response = openai.ChatCompletion.create(
-      engine="gpt-35-turbo",
-      messages = [{"role":"system","content":"You are an AI assistant that helps people find information."},{"role":"user","content":yourquestion}],
-      temperature=atemprature,
-      max_tokens=800,
-      top_p=1,
-      frequency_penalty=0,
-      presence_penalty=0,
-      stop=None)
+aimethod = st.radio(
+    "Choose a AI brain or document comprehension method",
+    ('2 minutes AI method map_reduce', '4 minutes AI method refine' ), index=0)
 
-    # Run the chain only specifying the input variable.
-    st.write(response)
+mychain_type = "map_reduce"
+if aimethod ==  '2 minutes AI method map_reduce':
+  mychain_type = "map_reduce"
+if aimethod ==  '4 minutes AI method refine':
+  mychain_type = "refine"
 
+chain = load_qa_chain(llm=llmgpt3, chain_type=mychain_type)
 
-if st.button("Ask Questions Simplify "):
-    template = """
-    You are an AI assistant. 
-    {concept}
-    """
 
-    response = openai.ChatCompletion.create(
-      engine="gpt-35-turbo",
-      messages = [{"role":"system","content":"You are an AI assistant that helps people find information. Please explain the information like i am a five."},{"role":"user","content":yourquestion}],
-      temperature=0,
-      max_tokens=800,
-      top_p=1,
-      frequency_penalty=0,
-      presence_penalty=0,
-      stop=None)
+loader1 = PyPDFLoader(url)
 
-    # Run the chain only specifying the input variable.
-    st.write(response)
 
-if st.button("Ask trying here "):
-    template = """
-    You are an expert on topics of Sustainability, Climate action and UN Sustainable Development Goals. 
-    Explain the concept of {concept}  like i am a five 
-    """
+def history():
+  mycount = 0 
+  if 'count' not in st.session_state:
+    st.session_state['count'] =  0
+  else:
+    mycount = st.session_state['count']
 
-    prompt = PromptTemplate(
-        input_variables=["concept"],
-        template=template,
-    )
+  if True:
+    st.write(mycount)
+    mycount = mycount + 1 
+    st.session_state['count'] = mycount
+    for i in range(mycount):
+      mystatekeyindex = "element" + str(i)
+      mystatekeyanswerindex = "elementANS" + str(i)
+      if mystatekeyindex not in st.session_state:
+        st.session_state[mystatekeyindex] =  yourquestion
+        st.session_state[mystatekeyanswerindex] = answer
+      if mystatekeyindex in st.session_state:
+        with st.expander(st.session_state[mystatekeyindex]):
+          st.write( st.session_state[mystatekeyanswerindex] )
 
 
-    from langchain.chains import LLMChain
-    chain = LLMChain(llm=llm, prompt=prompt)
+def colorizedtext(acolor, astring):
+  formattedcolor = ":" + acolor + "[" + astring +  "]"
+  return formattedcolor
 
-    # Run the chain only specifying the input variable.
-    st.write(chain.run(yourquestion))
 
+if st.button("Ask QA "):
+  documents = loader1.load()
+  answer = ""
+  with st.spinner(" Finding answer for your question "  ):
+    st.write("AI is reading this [link](%s)" % url)
+    with st.expander( "Employing your choice of AI method ... " + aimethod + "..."):
+      st.write(str(chain)[:100])
+    st.subheader(colorizedtext("red", yourquestion))
 
 
-if st.button("Ask Hindi "):
-    template = """
-    You are an expert on topics of Sustainability, Climate action and UN Sustainable Development Goals. 
-    Explain the concept of {concept} in Hindi
-    """
+    answer = chain.run(input_documents=documents, question=yourquestion)
+  st.subheader(colorizedtext("blue", answer))
+  history()
 
-    prompt = PromptTemplate(
-        input_variables=["concept"],
-        template=template,
-    )
 
 
-    from langchain.chains import LLMChain
-    chain = LLMChain(llm=llm, prompt=prompt)
 
-    # Run the chain only specifying the input variable.
-    st.write(chain.run(yourquestion))
+