Update app.py

app.py

@@ -1,80 +1,50 @@
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
-import openai
-import pinecone
-import torch
-import langchain as lc
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.vectorstores import Pinecone
-from langchain.chains import ConversationalRetrievalChain
-from langchain.agents import initialize_agent, load_tools
-from google.generativeai import genai
+import google.generativeai as genai
 
-# Configure Pinecone and API keys
-pinecone.init(api_key="YOUR_PINECONE_API_KEY", environment="YOUR_PINECONE_ENV")
+# Configure Google Gemini API
+genai.configure(api_key="AIzaSyDBGF5y7WqDB0SO7-UO6yjshiEZN3Hpt3g")  # Replace with your actual API key
 
-# Whisper model for audio transcription
-whisper_model = pipeline("automatic-speech-recognition", model="openai/whisper-large")
+# Function to get a response from the Google Gemini model
+def get_gemini_response(input_text):
+    # Check if the input is detailed enough
+    if len(input_text.split()) < 10:
+        return "Please provide a more detailed user story to help generate relevant needs and wants."
 
-# LLaMA 3.2 model for text processing
-llama_model_id = "meta-llama/Llama-3.2-3B-Instruct"
-tokenizer = AutoTokenizer.from_pretrained(llama_model_id)
-llama_model = AutoModelForCausalLM.from_pretrained(llama_model_id, torch_dtype=torch.bfloat16)
-
-# Embeddings for vector storage
-embedding_model = HuggingFaceEmbeddings("sentence-transformers/all-MiniLM-L6-v2")
-
-# Pinecone vector database setup
-pinecone_index = pinecone.Index("your-index-name")
-vector_db = Pinecone(index=pinecone_index, embedding_function=embedding_model.embed_query, text_key="text")
-
-# Define RAG chain using LangChain's ConversationalRetrievalChain
-rag_chain = ConversationalRetrievalChain.from_llm(
-    llm=llama_model,
-    retriever=vector_db.as_retriever(),
-    memory=lc.memory.ConversationBufferMemory()
-)
-
-# Function for Transcription, RAG, and Multilingual Response
-def process_input(input_text=None, input_audio=None):
-    # Step 1: Process Audio Input (if any) with Whisper
-    if input_audio:
-        transcription = whisper_model(input_audio)["text"]
-        input_text = transcription  # Use transcription as input for LLaMA 3.2 processing
+    # Concise prompt to limit output to essential insights
+    input_prompt = f"""
+    Based on the user story "{input_text}", briefly extract any unarticulated needs and wants.
     
-    # Step 2: Ensure there's text to process
-    if not input_text:
-        return "Please provide either a user story or an audio file."
-
-    # Step 3: Pass text to LLaMA 3.2 for primary processing
-    prompt = f"""
-    From the following user story, extract the unarticulated needs and wants.
-    User story: {input_text}
-
-    Needs are unspoken requirements or desires, while wants are explicit wishes.
+    Only provide essential needs and wants directly relevant to the given story. Do not speculate or over-extrapolate.
+    
+    Needs and Wants:
     """
-    inputs = tokenizer(prompt, return_tensors="pt")
-    llama_output = llama_model.generate(**inputs, max_new_tokens=100)
-    llama_response = tokenizer.decode(llama_output[0], skip_special_tokens=True)
 
-    # Step 4: RAG - Retrieve relevant documents from Pinecone and combine with LLaMA output
-    response_with_rag = rag_chain.run({"query": llama_response})
+    # Generate the content based on text input
+    model = genai.GenerativeModel('gemini-1.5-flash')
+    response = model.generate_content([input_text, input_prompt])
+    return response.text
 
-    return response_with_rag
+# Gradio interface function
+def extract_needs_and_wants(user_story):
+    try:
+        return get_gemini_response(user_story)
+    except Exception as e:
+        return f"Error: {str(e)}"
 
-# Gradio Interface for the application
-with gr.Blocks() as interface:
-    gr.Markdown("# Multimodal Needs & Wants Extractor with LLaMA 3.2 and Whisper")
-    input_text = gr.Textbox(label="User Story (Text Input)", placeholder="Enter your user story here...")
-    input_audio = gr.Audio(source="microphone", type="filepath", label="User Story (Audio Input)")
-    output_text = gr.Textbox(label="Response", placeholder="Extracted needs and wants will appear here...")
-    submit = gr.Button("Process")
-    
-    submit.click(
-        fn=process_input,
-        inputs=[input_text, input_audio],
-        outputs=output_text
-    )
+# Create the Gradio interface
+import gradio as gr
+
+interface = gr.Interface(
+    fn=extract_needs_and_wants,
+    inputs="text",
+    outputs="text",
+    title="Unarticulated Needs & Wants Extractor",
+    description="**Author:** VictorDaniel\n\nEnter a detailed user story to extract the unarticulated needs and wants.",
+    examples=[["The user often speaks about wanting to improve their health but is hesitant to join a gym."]]
+)
 
-# Launch the Gradio interface
 interface.launch()
+
+
+# Launch the Gradio app
+interface.launch()