Update README.md

README.md

@@ -20,6 +20,9 @@ State-of-the-art lightweights pretrained Transformer-based encoder-decoder model
 
 Model trained on dataset CNN-DailyMail News with input length = 512, output length = 150
 ## How to use
+Input for model: prefix + input text
+Example: 'summarize: '+ 'Ever noticed how plane seats.....'
+
 
 ```python
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM