Update app.py

app.py

@@ -37,10 +37,10 @@ warnings.filterwarnings("ignore")
 print("⚡ Chargement du modèle avec llama.cpp...")
 llm = Llama(
     model_path=MODEL_PATH,
-    n_ctx=4096,
+    n_ctx=2028,
     n_gpu_layers=0,
     n_threads=8,
-    n_batch=512,
+    #n_batch=512,
     verbose=False,
     use_mlock=True
 )
@@ -111,7 +111,7 @@ def send_message_stream(user_message, displayed_history, current_chat_name):
         stream = llm.create_completion(
             prompt=formatted_prompt,
             stream=True,
-            max_tokens=1024,
+            max_tokens=2028,
             temperature=0.7,
             top_p=0.9,
             repeat_penalty=1.1,