That’s how it does work, but unfortunately denoising the last paragraph requires computing attention scores for every token in that paragraph, which requires checking those tokens against every token in the sequence. So it’s still much less cacheable than the equivalent autoregressive model.
That’s how it does work, but unfortunately denoising the last paragraph requires computing attention scores for every token in that paragraph, which requires checking those tokens against every token in the sequence. So it’s still much less cacheable than the equivalent autoregressive model.