Update README.md

README.md

@@ -62,14 +62,21 @@ SteamSHP gets an average 72.8% accuracy across all domains:
 
 ## Usage
 
-Here's how to load the model:
+Here's how to use the model:
 
 ```python
 
-from transformers import T5ForConditionalGeneration, T5Tokenizer
+>> from transformers import T5ForConditionalGeneration, T5Tokenizer
+>> device = 'cuda'
 
-tokenizer = T5Tokenizer.from_pretrained('stanfordnlp/SteamSHP-preference-model')
-model = T5ForConditionalGeneration.from_pretrained('stanfordnlp/SteamSHP-preference-model')
+>> tokenizer = T5Tokenizer.from_pretrained('stanfordnlp/SteamSHP-preference-model')
+>> model = T5ForConditionalGeneration.from_pretrained('stanfordnlp/SteamSHP-preference-model').to(device)
+
+>> input_text = "POST: Instacart gave me 50 pounds of limes instead of 5 pounds... what the hell do I do with 50 pounds of limes? I've already donated a bunch and gave a bunch away. I'm planning on making a bunch of lime-themed cocktails, but... jeez. Ceviche? \n\n RESPONSE A: Lime juice, and zest, then freeze in small quantities.\n\n RESPONSE B: Lime marmalade lol\n\n Which response is better? RESPONSE"
+>> x = tokenizer([input_text], return_tensors='pt').input_ids.to(device)
+>> y = model.generate(x)
+>> tokenizer.batch_decode(y, skip_special_tokens=True)
+'A'
 ```
 
 The input text should be of the format: