Update app.py

app.py

@@ -12,7 +12,7 @@ client = InferenceClient(api_key=os.environ.get('HF_TOKEN'))
 # Constants for EOU calculation
 PUNCS = string.punctuation.replace("'", "")
 MAX_HISTORY = 4
-MAX_HISTORY_TOKENS = 512
+MAX_HISTORY_TOKENS = 1024
 EOU_THRESHOLD = 0.5
 
 # Initialize tokenizer and ONNX session
@@ -85,9 +85,9 @@ def chatbot(user_input):
     stream = client.chat.completions.create(
         model="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
         messages=messages,
-        temperature=0.5,
-        max_tokens=512,
-        top_p=0.7,
+        temperature=0.95,
+        max_tokens=1024,
+        top_p=0.6,
         stream=True
     )