Update app.py

app.py

@@ -1,57 +1,14 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
+# Initialize the client
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value='''
-# Update MEDICAL_PROMPT to be more restrictive
-MEDICAL_PROMPT = PromptTemplate(
-    input_variables=["query"],
-    template="""<bos><start_of_turn>system
+# Define the medical system prompt
+MEDICAL_SYSTEM_PROMPT = """<bos><start_of_turn>system
 You are Gemma, a medical AI assistant. You MUST ONLY answer health and medical-related questions.
+USE BULLET POINTS TO STRUCTURE YOUR ANSWERS AND THEY MUST BE PUNCTUATED AND GRAMMATICALLY CORRECT.
+Your responses should be concise and informative.
 Your responses should be professional, accurate, and focused on medical topics only.
 For any non-medical questions, respond with a redirection to medical topics.
 For medication queries, provide general information and recommend consulting a healthcare professional.
@@ -190,54 +147,63 @@ IMPORTANT NOTES:
 <end_of_turn>
 <start_of_turn>user
 {query}<end_of_turn>
-<start_of_turn>model
-'''
-)
+<start_of_turn>model."""
 
-# Update is_medical_query to be more comprehensive
+# Function to check if a query is medical-related
 def is_medical_query(query):
-        medical_keywords_and_greetings = [
-            "health", "disease", "symptom", "doctor", "medicine", "medical", "treatment",
-            "hospital", "clinic", "diagnosis", "patient", "drug", "prescription", "therapy",
-            "cancer", "diabetes", "heart", "blood", "pain", "surgery", "vaccine", "infection",
-            "allergy", "diet", "nutrition", "vitamin", "exercise", "mental health", "depression",
-            "anxiety", "disorder", "syndrome", "chronic", "acute", "emergency", "pharmacy",
-            "dosage", "side effect", "contraindication", "body", "organ", "immune", "virus",
-            "bacterial", "fungal", "parasite", "genetic", "hereditary", "congenital", "prenatal",
-            "headaches", "ache", "stomach ache", "skin", "head", "arm", "leg", "chest", "back", "throat", "eye", "ear", "nose", "mouth"
-        ]
-        
-        # Remove greetings from the keyword list
-        medical_keywords = [word for word in medical_keywords_and_greetings if word not in ["hello", "hi", "greetings", "good morning", "good afternoon", "good evening", "hey"]]
-        
-        query_lower = query.lower()
-        return any(keyword in query_lower for keyword in medical_keywords)
-
-        
-# Update chat_with_model to enforce medical-only responses
-def chat_with_model(message, history):
+    medical_keywords = [
+        "health", "disease", "symptom", "doctor", "medicine", "medical", "treatment",
+        "hospital", "clinic", "diagnosis", "patient", "drug", "prescription", "therapy",
+        "cancer", "diabetes", "heart", "blood", "pain", "surgery", "vaccine", "infection",
+        "allergy", "diet", "nutrition", "vitamin", "exercise", "mental health", "depression",
+        "anxiety", "disorder", "syndrome", "chronic", "acute", "emergency", "pharmacy",
+        "dosage", "side effect", "contraindication", "body", "organ", "immune", "virus",
+        "bacterial", "fungal", "parasite", "genetic", "hereditary", "congenital", "prenatal",
+        "headaches", "ache", "stomach ache", "skin", "head", "arm", "leg", "chest", "back", 
+        "throat", "eye", "ear", "nose", "mouth"
+    ]
+    
+    query_lower = query.lower()
+    return any(keyword in query_lower for keyword in medical_keywords)
+
+# Response function for the chatbot
+def respond(message, history, max_tokens=512, temperature=0.7, top_p=0.95):
+    # Check if the query is medical-related
+    if not is_medical_query(message):
+        return "I'm specialized in medical topics only. I cannot answer this question. How can I assist with a health-related concern instead?"
+    
+    # Prepare the messages for the API
+    messages = [{"role": "system", "content": MEDICAL_SYSTEM_PROMPT}]
+    
+    for user_msg, assistant_msg in history:
+        if user_msg:
+            messages.append({"role": "user", "content": user_msg})
+        if assistant_msg:
+            messages.append({"role": "assistant", "content": assistant_msg})
+    
+    messages.append({"role": "user", "content": message})
+    
+    # Generate the response
+    response = ""
+    
     try:
-        context = "\n".join([f"User: {msg}\nAssistant: {res}" for msg, res in history])
-        full_query = f"{context}\nUser: {message}"
-        
-        if not is_medical_query(full_query):
-            return "I'm specialized in medical topics only. I cannot answer this question. How can I assist with a health-related concern instead?"
-        
-        response = medical_chain.run(query=full_query)
-        clean_response = response.split("<end_of_turn>")[0].strip()
-        
-        # Check if the response is medical-related
-        if not is_medical_query(clean_response):
-            return "I can only provide information on medical topics. Please ask a medical question."
-        
-        return clean_response
-        
+        for message in client.chat_completion(
+            messages,
+            max_tokens=max_tokens,
+            stream=True,
+            temperature=temperature,
+            top_p=top_p,
+        ):
+            token = message.choices[0].delta.content
+            if token:
+                response += token
+                yield response
     except Exception as e:
-        return f"I apologize, but I encountered an error: {str(e)}. Please try again."
+        yield f"I apologize, but I encountered an error: {str(e)}. Please try again."
 
-# Update Gradio examples to be medical-specific
-iface = gr.ChatInterface(
-    fn=chat_with_model,
+# Create the Gradio interface
+demo = gr.ChatInterface(
+    fn=respond,
     title="MedexDroid - Medical Assistant",
     examples=[
         "What are the symptoms of diabetes?",
@@ -249,19 +215,8 @@ iface = gr.ChatInterface(
     description="An AI Medical Assistant. Please ask health-related questions only.",
     theme=gr.themes.Soft(),
     css=".gradio-container {background-color: #f0f4f8}"
-, label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
 )
 
-
 if __name__ == "__main__":
     demo.launch()
+