DiffusionBlocks: Training Neural Networks One Block at a Time

4 points by sebg 28 days ago | 1 comment

billconan 28 days ago |

I do not understand.

how is this different from building smaller transformer layers, and each layer just denoises less?

billconan 28 days ago |

I do not understand.

how is this different from building smaller transformer layers, and each layer just denoises less?