Update app.py

app.py

@@ -41,12 +41,13 @@ def model_to_cuda(model):
     return model
 
 def flan(model_name, sl, tl, input_text):
-    tokenizer = T5Tokenizer.from_pretrained(model_name)
+    tokenizer = T5Tokenizer.from_pretrained(model_name, legacy=False)
     model = T5ForConditionalGeneration.from_pretrained(model_name)
     input_text = f"translate {sl} to {tl}: {input_text}?"
     input_ids = tokenizer(input_text, return_tensors="pt").input_ids
     outputs = model.generate(input_ids)
-    return tokenizer.decode(outputs[0])
+    return tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
+    # return tokenizer.decode(outputs[0], skip_special_tokens=True).replace('<pad>', '').replace('</s>', '').strip()
 
 def teuken(model_name, sl, tl, input_text):
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")