Create added_tokens.json

Copied from alpaca-13B and native, solves an error message when converting to ggml files

Exception: Vocab size mismatch (model has 32001, but models/chavinlo_gpt4-x-alpaca/tokenizer.model has 32000)

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "[PAD]": 32000
+}