update generation config to defaults similar to alpaca-lora

generation_config.json

@@ -3,5 +3,10 @@
   "bos_token_id": 0,
   "eos_token_id": 0,
   "transformers_version": "4.28.1",
-  "use_cache": false
-}
+  "use_cache": false,
+  "temperature": 0.1,
+  "top_p": 0.75,
+  "top_k": 40,
+  "num_beams": 4,
+  "max_new_tokens": 128
+}