Upload app.py

app.py

@@ -3,18 +3,46 @@
 ## Setup
 # Import the necessary Libraries
 
+import json
+import tiktoken
+import gradio as gr
+import uuid
 
+import pandas as pd
+
+from openai import OpenAI
+
+from langchain_community.embeddings.sentence_transformer import (
+    SentenceTransformerEmbeddings
+)
+from langchain_community.vectorstores import Chroma
+
+from google.colab import userdata, drive
+from huggingface_hub import CommitScheduler
 
 
 
 
 # Create Client
+load_dotenv()
 
+os.environ["ANYSCALE_API_KEY"]=os.getenv("ANYSCALE_API_KEY")
+
+client = OpenAI(
+    base_url="https://api.endpoints.anyscale.com/v1",
+    api_key=os.environ['ANYSCALE_API_KEY']
+)
 
 # Define the embedding model and the vectorstore
+embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')
 
 # Load the persisted vectorDB
 
+reportdb = Chroma(
+    collection_name=collection_name,
+    persist_directory='./report_db1',
+    embedding_function=embedding_model
+)
 
 # Prepare the logging functionality
 
@@ -22,7 +50,7 @@ log_file = Path("logs/") / f"data_{uuid.uuid4()}.json"
 log_folder = log_file.parent
 
 scheduler = CommitScheduler(
-    repo_id="---------",
+    repo_id="report-logs",
     repo_type="dataset",
     folder_path=log_folder,
     path_in_repo="data",
@@ -31,12 +59,30 @@ scheduler = CommitScheduler(
 
 # Define the Q&A system message
 
+qna_system_message = """
+You are an assistant to a financial services firm who answers user queries on annual reports.
+User input will have the context required by you to answer user questions.
+This context will begin with the token: ###Context.
+The context contains references to specific portions of a document relevant to the user query.
+
+User questions will begin with the token: ###Question.
 
+Please answer only using the context provided in the input. Do not mention anything about the context in your final answer.
+
+If the answer is not found in the context, respond "I don't know".
+"""
 
 
 
 # Define the user message template
+qna_user_message_template = """
+###Context
+Here are some documents that are relevant to the question mentioned below.
+{context}
 
+###Question
+{question}
+"""
 
 
 
@@ -48,12 +94,33 @@ def predict(user_input,company):
 
     # Create context_for_query
 
+relevant_document_chunks = retriever.get_relevant_documents(user_question)
+context_list = [d.page_content for d in relevant_document_chunks]
+context_for_query = ". ".join(context_list)
 
-    # Create messages
+prompt = [
+    {'role':'system', 'content': qna_system_message},
+    {'role': 'user', 'content': qna_user_message_template.format(
+         context=context_for_query,
+         question=user_question
+        )
+    }
+]
 
 
-    # Get response from the LLM
 
+print(prompt)
+
+    # Create messages
+response = client.chat.completions.create(
+    model=model_name,
+    messages=prompt,
+    temperature=0
+)
+
+    # Get response from the LLM
+answer = response.choices[0].message.content.strip()
+print (answer)
 
     # While the prediction is made, log both the inputs and outputs to a local log file
     # While writing to the log file, ensure that the commit scheduler is locked to avoid parallel
@@ -82,6 +149,7 @@ company = gr.Radio()
 # Create the interface
 # For the inputs parameter of Interface provide [textbox,company]
 
+demo = gr.Interface(inputs=[textbox,company], fn = predict, output ='text')
 
 demo.queue()
 demo.launch()