Test repo to experiment with calling generate from the hub. It is a simplified implementation of greedy decoding.
generate
left_padding (int, optional): number of padding tokens to add before the provided input
left_padding
int
(none)
-