mejoras 2

app.py

@@ -23,11 +23,10 @@ def respond(message: str, history: List[Tuple[str, str]], system_message: str, m
         messages,
         max_tokens=max_tokens,
         stream=True,
-        temperature=temperature,
-        top_p=top_p,
-        repetition_penalty=1.2,  # Add repetition penalty
-        presence_penalty=0.5,    # Penalize presence of repeated tokens
-        frequency_penalty=0.5,   # Penalize frequency of repeated tokens
+        temperature=0.7,        # Aumentado para más variedad
+        top_p=0.85,            # Ajustado para mejor balance
+        top_k=40,              # Limita las opciones a las 40 mejores
+        do_sample=True,        # Habilita muestreo aleatorio
     ):
         token = msg.choices[0].delta.content
         response += token
@@ -36,10 +35,17 @@ def respond(message: str, history: List[Tuple[str, str]], system_message: str, m
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=256, 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)"),
+        gr.Textbox(
+            value="""You are a helpful assistant. Follow these rules:
+1. Provide diverse and varied responses
+2. Avoid repeating the same words or phrases
+3. Use synonyms and alternative expressions
+4. Be concise and direct""",
+            label="System message"
+        ),
+        gr.Slider(minimum=1, maximum=256, value=200, step=1, label="Max new tokens"),
+        gr.Slider(minimum=0.7, maximum=1.2, value=1.0, step=0.1, label="Temperature"),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.85, step=0.05, label="Top-p (nucleus sampling)"),
     ],
 )