Update README.md

According to https://github.com/facebookresearch/fairseq, "fairseq(-py) is MIT-licensed. The license applies to the pre-trained models as well." Therefore, as mBART is explicitly stated as a fine-tune, it is tied to Fairseq (see Fairseq repository), and inherits the MIT license.

README.md

@@ -56,6 +56,7 @@ language:
 tags:
 - mbart-50
 pipeline_tag: translation
+license: mit
 ---
 
 # mBART-50 many to many multilingual machine translation
@@ -115,4 +116,4 @@ Arabic (ar_AR), Czech (cs_CZ), German (de_DE), English (en_XX), Spanish (es_XX),
     archivePrefix={arXiv},
     primaryClass={cs.CL}
 }
-```
+```