Update app.py

app.py

@@ -29,16 +29,15 @@ SYSTEM_PROMPT = (
 )
 
 def build_context(user_message):
-    # Stateless: only system prompt + current message
     return SYSTEM_PROMPT + "\n\nUser: " + user_message + "\nAssistant:"
 
 def generate_response(
     prompt,
-    max_tokens=150,
-    temperature=0.5,
-    top_p=0.92,
-    repetition_penalty=1.08,
-    top_k=40,
+    max_tokens=300,
+    temperature=0.45,
+    top_p=0.95,
+    repetition_penalty=1.1,
+    top_k=35,
 ):
     formatted_input = build_context(prompt)
 
@@ -92,11 +91,11 @@ chatbot = gr.ChatInterface(
     title="Chat with source-mini-beta",
     description="Open-source chatbot, beta development, 0 restrictions, answers all topics.",
     additional_inputs=[
-        gr.Slider(minimum=10, maximum=200, value=150, step=5, label="Max new tokens"),
-        gr.Slider(minimum=0.01, maximum=1.0, value=0.5, step=0.01, label="Temperature"),
-        gr.Slider(minimum=0.5, maximum=1.0, value=0.92, step=0.01, label="Top-p (nucleus sampling)"),
-        gr.Slider(minimum=1.0, maximum=1.5, value=1.08, step=0.001, label="Repetition penalty"),
-        gr.Slider(minimum=1, maximum=100, value=40, step=1, label="Top-k (prediction sampling)"),
+        gr.Slider(minimum=50, maximum=500, value=300, step=10, label="Max new tokens"),
+        gr.Slider(minimum=0.01, maximum=1.0, value=0.45, step=0.01, label="Temperature"),
+        gr.Slider(minimum=0.5, maximum=1.0, value=0.95, step=0.01, label="Top-p (nucleus sampling)"),
+        gr.Slider(minimum=1.0, maximum=1.5, value=1.1, step=0.001, label="Repetition penalty"),
+        gr.Slider(minimum=1, maximum=100, value=35, step=1, label="Top-k (prediction sampling)"),
     ],
 )