Updated LLM generation parameters

app.py

@@ -43,9 +43,9 @@ def generate_llm_response(text, model_id="ccibeekeoc42/Llama3.1-8b-base-SFT-2024
                 {"role": "system", "content": "You are a BRIEF AND DIRECT assistant. A part of a speech pipeline so keep your responses short, fluent, and straight to the point. Avoid markdown in responses."},
                 {"role": "user", "content": text}
             ],
-            top_p=None,
-            temperature=None,
-            max_tokens=75,
+            top_p=0.3,
+            temperature=1,
+            max_tokens=750,
             stream=True,
             seed=None,
             stop=None,