new trained data

This is freshly trained on C4 subset as well as Daily Dialog and AAC text and more..

tokenizer_config.json

@@ -930,8 +930,10 @@
   "clean_up_tokenization_spaces": true,
   "eos_token": "</s>",
   "extra_ids": 100,
+  "legacy": false,
   "model_max_length": 512,
   "pad_token": "<pad>",
+  "sp_model_kwargs": {},
   "tokenizer_class": "T5Tokenizer",
   "unk_token": "<unk>"
 }