bigscience

/

mt0-base

like 29

I am new to t5 and mt5. I found the vocab_size in tokenizer is 250100, but the model's embedding size is 250112. Could you tell me why they are different?
Thx in advance~