Update README.md

README.md

@@ -20,10 +20,11 @@ DEVICE = "cuda:0" if torch.cuda.is_available() else "cpu"
 PATH = "josh-oo/aspect-based-embeddings-v3"
 
 tokenizer = AutoTokenizer.from_pretrained(PATH)
+tokenizer.model_max_length = 512
 model = AutoModel.from_pretrained(PATH)
 
 dummy_text = "This is a title of a medical paper"
-dummy_input = tokenizer([dummy_text], return_tensors="pt")
+dummy_input = tokenizer([dummy_text], return_tensors="pt", truncation=True)
 
 dummy_input.to(DEVICE)
 model.to(DEVICE)
@@ -54,7 +55,7 @@ model.register_buffer("token_type_ids", torch.zeros(model.position_ids.size(), d
 model.token_type_ids[:,1:1+len(PREFIXES)] = 1 #set prefix token type ids to 1
 
 dummy_text = "".join(PREFIXES) + "This is a title of a medical paper"
-dummy_input = tokenizer([dummy_text], return_tensors="pt")
+dummy_input = tokenizer([dummy_text], return_tensors="pt", truncation=True)
 dummy_input.pop('token_type_ids')
 
 dummy_input.to(DEVICE)