Update app.py

app.py

@@ -3,14 +3,25 @@ from llama_cpp import Llama
 
 llm = Llama(
     model_path="yugogpt-q4_k_s.gguf",
-    n_ctx=4096
+    n_ctx=4096,            # Doubled context length
+    n_threads=8,           # Increased threads
+    n_batch=1024,          # Increased batch size
+    use_mlock=True,
+    use_mmap=True,
+    n_gpu_layers=0,        # Set this to higher number if GPU available
+    verbose=False          # Reduced logging for better performance
 )
 
 def chat(message, history):
     response = llm.create_completion(
         f"USER: {message}\nASSISTANT:",
-        max_tokens=512,
-        temperature=0.7
+         max_tokens=4096,    # Increased max tokens
+        temperature=0.1,    # Keeping it precise
+        top_p=0.1,
+        repeat_penalty=1.2,
+        top_k=20,
+        stop=["USER:", "\n\n"],
+        stream=True
     )
     return response['choices'][0]['text']