open ai support

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 from datetime import datetime, timedelta
 
 from gemini_api import model_api, sentiment, category, ord_num, NO_ORDER, NO_ITEM, food_return, cloth_return, item_identy, item_match
+from openai_api import model_api as openai_model_api, sentiment as openai_sentiment
 
 cust_qry_resp = {"senti":"", "cat":"", "num":""}
 
@@ -17,6 +18,7 @@ with gr.Blocks(title="Customer Support Assistant",
   state_match_item = gr.State([None, None])
   
   gr.Markdown("# Customer Support Assistant")
+  llm_api = gr.Radio(["gemini-1.0-pro", "gpt-3.5-turbo"], label="Choose LLM", value="gemini-1.0-pro")
   # Inputs from user
   with gr.Row():
     cust_qry = gr.Textbox(lines=5, type="text", label="Customer Query")
@@ -24,13 +26,16 @@ with gr.Blocks(title="Customer Support Assistant",
   btn_cust_qry = gr.Button("Analyze Query")
 
   # Model Output
-  @gr.render(inputs=[cust_qry], triggers=[btn_cust_qry.click])
+  @gr.render(inputs=[cust_qry, llm_api], triggers=[btn_cust_qry.click])
   # Function for prediction
-  def invoke_model(user_input):
+  def invoke_model(user_input, llm_api):
     if len(user_input) == 0:
       gr.Markdown("## No Customer Query Provided")
     else:  
-      senti = model_api(user_input, sentiment)
+      if llm_api == "gemini-1.0-pro":
+        senti = model_api(user_input, sentiment)
+      else:
+        senti = openai_model_api(user_input, openai_sentiment)
       cat = model_api(user_input, category)
       num = model_api(user_input, ord_num)
       item = model_api(user_input, item_identy)

openai_api.py

@@ -1,6 +1,5 @@
 import openai
 from openai import OpenAI
-import json
 import os
 
 # Set your OpenAI API key (replace with your actual key)
@@ -13,10 +12,11 @@ def model_api(input, prompt_type):
     return prompt_type(input)
 
 def sentiment(text):
+    print(text)
     # Create a prompt for the model
     prompt = f"""You are trained to analyze and detect the sentiment of the given text.
     If you are unsure of an answer, you can say "not sure" and recommend the user review manually.
-    Analyze the following text and determine if the sentiment is: Positive, Negative, or Neutral.
+    Analyze the following text and determine if the sentiment is: POSITIVE, NEGATIVE, NEUTRAL or MIXED.
     {text}"""
 
     # Call the OpenAI API to generate a response
@@ -29,7 +29,7 @@ def sentiment(text):
         max_tokens=1,  # Limit response to a single word
         temperature=0  # Keep response consistent
     )
-
+    print(response)
     # Extract the sentiment from the response
     sentiment = response.choices[0].message.content.strip().lower()
 

requirements.txt

@@ -1,2 +1,3 @@
 gradio
-google-generativeai
+google-generativeai
+langchain_openai