Update app.py

app.py

@@ -18,15 +18,16 @@ Does the human question violate the above policy? Your answer must start with 'Y
     return prompt
 
 def verificar_politica(message):
-    policies = [
-        "No Dangerous Content: The prompt shall not contain or seek generation of content that harms oneself and/or others (e.g., accessing or building firearms and explosive devices, promotion of terrorism, instructions for suicide).",
-        "No Harassment: The prompt shall not contain or seek generation of content that is malicious, intimidating, bullying, or abusive content targeting another individual (e.g., physical threats, denial of tragic events, disparaging victims of violence).",
-        "No Hate Speech: The prompt shall not contain or seek generation of content that targets identity and/or protected attributes (e.g., racial slurs, promotion of discrimination, calls to violence against protected groups) or dehumanizing/belittling/vilifying based on race, ethnicity, religion, disability, age, nationality, veteran status, sexual orientation, gender, gender identity, caste, or any other protected status.",
-        "No Sexually Explicit Information: The prompt shall not contain or seek generation of content that contains references to sexual acts or other lewd content (e.g., sexually graphic descriptions, content aimed at causing arousal). Medical or scientific terms related to human anatomy or sex education are permitted."
-    ]
+    policies = {
+        "No Dangerous Content": "The prompt shall not contain or seek generation of content that harms oneself and/or others (e.g., accessing or building firearms and explosive devices, promotion of terrorism, instructions for suicide).",
+        "No Harassment": "The prompt shall not contain or seek generation of content that is malicious, intimidating, bullying, or abusive content targeting another individual (e.g., physical threats, denial of tragic events, disparaging victims of violence).",
+        "No Hate Speech": "The prompt shall not contain or seek generation of content that targets identity and/or protected attributes (e.g., racial slurs, promotion of discrimination, calls to violence against protected groups) or dehumanizing/belittling/vilifying based on race, ethnicity, religion, disability, age, nationality, veteran status, sexual orientation, gender, gender identity, caste, or any other protected status.",
+        "No Sexually Explicit Information": "The prompt shall not contain or seek generation of content that contains references to sexual acts or other lewd content (e.g., sexually graphic descriptions, content aimed at causing arousal). Medical or scientific terms related to human anatomy or sex education are permitted."
+    }
 
-    for policy in policies:
-        prompt = generar_prompt(message, policy)
+    scores = {}
+    for policy_name, policy_text in policies.items():
+        prompt = generar_prompt(message, policy_text)
         response = llm.create_chat_completion(
             messages=[{"role": "system", "content": prompt}],
             stream=False
@@ -34,15 +35,23 @@ def verificar_politica(message):
         answer = response['choices'][0]['message']['content']
         
         # Calcula las probabilidades de "Yes" y "No" basándote en la respuesta generada
-        yes_probability = answer.lower().count("yes") / (answer.lower().count("yes") + answer.lower().count("no"))
-        no_probability = 1 - yes_probability
-
-        print(f"Policy: {policy}")
-        print(f"Yes probability: {yes_probability}")
-        print(f"No probability: {no_probability}")
+        yes_count = answer.lower().count("yes")
+        no_count = answer.lower().count("no")
+        total_count = yes_count + no_count
+        
+        if total_count == 0:
+            yes_probability = 0
+        else:
+            yes_probability = yes_count / total_count
+        
+        scores[policy_name] = yes_probability
+        print(f"Policy: {policy_name}")
+        print(f"Score: {yes_probability:.2f}")
 
-        if yes_probability > no_probability:
-            return f"Your question violates the following policy: {policy.split(':')[0]}"
+    # Determina si alguna política tiene una puntuación alta
+    for policy_name, score in scores.items():
+        if score > 0.5:
+            return f"Your question violates the following policy: {policy_name} with a score of {score:.2f}"
     return None
 
 def respond(message, history, system_prompt):
@@ -79,7 +88,26 @@ def chat_stream_completion(message, history, system_prompt):
             message_repl += chunk['choices'][0]["delta"]["content"]
             yield message_repl
 
-gr.ChatInterface(
-    chat_stream_completion,
-    additional_inputs=[gr.Textbox("You are a helpful AI.", label="System Prompt")]
-).queue().launch(server_name="0.0.0.0")
+# Interfaz de Gradio con la descripción y configuración adicional
+with gr.Blocks() as demo:
+    gr.Markdown("# Child-Safe-Chatbot (Experimental)")
+    gr.Markdown("""
+    ### Description
+    This chatbot is designed to assist users while ensuring that all interactions comply with defined safety policies. It checks whether user inputs violate any of the following categories:
+    
+    - Dangerous Content
+    - Harassment
+    - Hate Speech
+    - Sexually Explicit Information
+    
+    The chatbot will inform the user if any violation occurs and, if not, will proceed to respond to the user's message in a friendly manner.
+    
+    I'm Norberto Martín Afonso. Follow me on [Twitter](https://twitter.com/norbertomartnaf) and [GitHub](https://github.com/nmarafo) for more updates and projects!
+    """)
+    
+    gr.ChatInterface(
+        chat_stream_completion,
+        additional_inputs=[gr.Textbox("You are a helpful AI.", label="System Prompt")]
+    ).queue().launch(server_name="0.0.0.0")
+
+demo.launch(debug=True)