Update app.py

app.py

@@ -9,7 +9,6 @@ import json
 import uuid
 import tiktoken
 from openai import OpenAI
-#from dotenv import load_dotenv
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_core.documents import Document
 from langchain_community.document_loaders import PyPDFDirectoryLoader
@@ -17,16 +16,11 @@ from langchain_community.embeddings.sentence_transformer import (
     SentenceTransformerEmbeddings
 )
 from langchain_community.vectorstores import Chroma
-
 from langchain_community.chat_models import ChatOpenAI
-
 from huggingface_hub import CommitScheduler
 from pathlib import Path
 
-
-
 # Create Client
-#anyscale_api_key = 'esecret_gseacat1ltrafee81njti6867e'
 anyscale_api_key = os.getenv('anyscale_apiKey')
 
 client = OpenAI(base_url="https://api.endpoints.anyscale.com/v1",
@@ -34,8 +28,6 @@ client = OpenAI(base_url="https://api.endpoints.anyscale.com/v1",
 )
 
 # Define the embedding model and the vectorstore
-#model_name = "thenlper/gte-large"
-#embedding_model = "thenlper/gte-large"
 model_name = "mlabonne/NeuralHermes-2.5-Mistral-7B"
 embedding_model = SentenceTransformerEmbeddings(model_name="thenlper/gte-large")
 
@@ -49,9 +41,8 @@ vectorstore_persisted = Chroma(
     persist_directory=persisted_vectordb_location,
     embedding_function=embedding_model
 )
-test = str(vectorstore_persisted.get())[:20]
-# Prepare the logging functionality
 
+# Prepare the logging functionality
 log_file = Path("logs/") / f"data_{uuid.uuid4()}.json"
 log_folder = log_file.parent
 
@@ -143,8 +134,7 @@ def predict(user_input,company):
      )
          
       prediction = response.choices[0].message.content.strip() 
-      #prediction =  "CONTEXT LIST: " + str(context_list) + "CONTEXT FOR QUERY:" + str(context_for_query)  
-      #prediction = test + ' length of chunks (should equal 5) ' + str(len(relevant_document_chunks))    
+        
     except Exception as e:
       prediction = f'Sorry, I encountered the following error: \n {e}'
 
@@ -175,11 +165,8 @@ def predict(user_input,company):
 # The submit button is used to run the predict function
 
 
-textbox = gr.Textbox()
-company = gr.Radio(choices=['aws', 'google', 'Meta', 'msft', 'IBM'], label="Select a company:")
-   
-#predict = gr.Button("Submit")
-#predict.click(predict, inputs=[textbox,company], outputs=[predict])
+textbox = gr.Textbox(label="Enter your question:")
+company = gr.Radio(choices=['aws', 'google', 'Meta', 'msft', 'IBM'], label="Select a company:")  
 
 
 model_output = gr.Label(label='Answer to your qestion')