model controllability

pretraining objective (1) encoder-decoder language model (1) zero-shot control (1) downstream controllability (1) encoder-decoder model (1) denoising objective (1) sequence-to-sequence model (1) language modeling (1) self-supervised learning (1)