File size: 3,401 Bytes
3b07bbb | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 | import gradio as gr
from huggingface_hub import InferenceClient
from sentence_transformers import SentenceTransformer
import torch
import re
# Initialize models and client
embedding_model = SentenceTransformer('all-MiniLM-L6-v2')
client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.2")
# Load Medicaid/health info
with open("medicaid_info.txt", "r", encoding="utf-8") as file:
health_info = file.read()
# Load clinic info
with open("info.txt", "r", encoding="utf-8") as file:
clinic_data = file.read().lower()
# Preprocess info into chunks
def preprocess_chunks(text):
chunks = [chunk.strip() for chunk in text.split('.') if chunk.strip()]
return chunks
health_chunks = preprocess_chunks(health_info)
health_embeddings = embedding_model.encode(health_chunks, convert_to_tensor=True)
# Semantic search
def get_relevant_chunks(query, embeddings, text_chunks):
query_embedding = embedding_model.encode(query, convert_to_tensor=True)
query_embedding = query_embedding / query_embedding.norm()
embeddings = embeddings / embeddings.norm(dim=1, keepdim=True)
similarities = torch.matmul(embeddings, query_embedding)
top_indices = torch.topk(similarities, k=3).indices
return [text_chunks[i] for i in top_indices]
# Clinic finder
def find_clinic_by_county(county):
county = county.lower()
sections = clinic_data.split("###")
for section in sections:
if county in section:
lines = section.strip().split("\n")
if lines and county in lines[0].lower():
lines = lines[1:]
return "\n".join(lines).strip()
return "⚠️ Sorry, I couldn’t find clinics for that county. Check spelling or try a nearby county."
# Chatbot logic
def respond(message, history, name, focus_area):
focus = focus_area[0] if focus_area else "general help"
top_chunks = get_relevant_chunks(message, health_embeddings, health_chunks)
context = "\n".join(top_chunks)
messages = [
{
"role": "system",
"content": (
f"You are a friendly and supportive health advisor chatbot helping Washington residents, especially low-income individuals. "
f"You're chatting with {name}. Speak clearly and kindly, and keep responses under 100 words. Focus on {focus}."
)
},
{
"role": "user",
"content": f"Context:\n{context}\n\nQuestion: {message}"
}
]
response = client.chat_completion(messages, max_tokens=120)
return response['choices'][0]['message']['content'].strip()
# Interface styling
title = "# 🩺 HealthHelpBot"
description = "### Helping Washington residents find care, understand Medicaid, and locate free clinics."
with gr.Blocks(theme="gstaff/soft-blue") as demo:
with gr.Row():
gr.Markdown(title)
with gr.Row():
gr.Markdown(description)
with gr.Row():
with gr.Column(scale=1):
gr.Image("healthbot_avatar.png", width=250, show_label=False)
user_name = gr.Textbox(label="👤 Your Name", placeholder="e.g. Maria")
focus_area = gr.CheckboxGroup(["Medicaid", "Finding a clinic", "Insurance help"], label="What do you need help with?")
with gr.Column(scale=2):
gr.ChatInterface(fn=respond, additional_inputs=[user_name, focus_area], type="messages")
demo.launch()
|