Encode separator for completions endpoint

`SEPARATOR` is being used to construct a prompt for use with the completions endpoint

examples/Question_answering_using_embeddings.ipynb

@@ -554,7 +554,7 @@
    "source": [
     "MAX_SECTION_LEN = 500\n",
     "SEPARATOR = \"\\n* \"\n",
-    "ENCODING = \"cl100k_base\"  # encoding for text-embedding-ada-002\n",
+    "ENCODING = \"gpt2\"  # encoding for text-davinci-003\n",
     "\n",
     "encoding = tiktoken.get_encoding(ENCODING)\n",
     "separator_len = len(encoding.encode(SEPARATOR))\n",