Update app.py

app.py

@@ -16,9 +16,13 @@ adapter_path = hf_hub_download(repo_id=adapter_repo, filename=adapter_file)
 llm = Llama(
     model_path=base_model_path,
     lora_path=adapter_path,
-    n_ctx=512,       # Context length, set manually since adapter lacks it
-    n_threads=2,     # Adjust based on your system
-    n_gpu_layers=0    # Set to >0 if GPU acceleration is desired and supported
+    n_ctx=256,          # Reduced context length (adjust based on your needs)
+    n_threads=8,        # Use all available CPU cores (adjust based on your system)
+    n_gpu_layers=0,    # Enable GPU acceleration (adjust based on VRAM)
+    use_mmap=False,     # Disable memory mapping for faster access
+    use_mlock=True,     # Lock model in memory for stability
+    logits_all=False,   # Only compute necessary logits
+    embedding=False    # Set to >0 if GPU acceleration is desired and supported
 )
 
 # Define the translation function