# app.py import pandas as pd import numpy as np import faiss from sentence_transformers import SentenceTransformer import gradio as gr # Load data data = pd.read_csv("Mental_Health_FAQ.csv") # Rename columns if necessary data.rename(columns={ 'Questions': 'question', 'Answers': 'answer' }, inplace=True) # Load model and build embeddings model = SentenceTransformer('all-MiniLM-L6-v2') faq_embeddings = np.array(model.encode(data['question'].tolist())) # Create FAISS index faq_index = faiss.IndexFlatL2(faq_embeddings.shape[1]) faq_index.add(faq_embeddings) # Define query function def gradio_faq(query): query_embedding = model.encode([query]) D, I = faq_index.search(np.array(query_embedding), 3) results = "" for rank, idx in enumerate(I[0]): question = data.iloc[idx]['question'] answer = data.iloc[idx]['answer'] results += f"🔹 **Q{rank+1}:** {question}\n**A:** {answer}\n\n" return results.strip() # Launch Gradio interface gr.Interface( fn=gradio_faq, inputs=gr.Textbox(label="Ask a Mental Health Question"), outputs=gr.Markdown(label="Top Answers"), title="🧠 Mental Health FAQ Assistant", description="Type your question to get the most relevant answers from the FAQ dataset." ).launch()