Update model_runner.py

model_runner.py

@@ -13,10 +13,9 @@ class ModelRunner:
         )
         self.llm = Llama(
             model_path=model_path,
-            flash_attn=True,
-            n_gpu_layers=81,
-            n_batch=1024,
-            n_ctx=8192,
+            n_gpu_layers=0,       # ⛔ No GPU
+            n_batch=512,          # safe default for CPU
+            n_ctx=4096            # lower context for CPU environments
         )
         self.provider = LlamaCppPythonProvider(self.llm)
         self.agent = LlamaCppAgent(