Increasing transformer token length with a Maximum Entropy Principle Method
This addresses efficiency issues for users of transformer models, though it appears incremental as it builds on existing methods with a new intermediate step.
The paper tackles the computational overhead of transformers due to quadratic sequence length dependence by introducing three methods based on the Maximum Entropy Principle to extend autoregressive token length from T to 2T tokens linearly, with some overhead but claimed faster performance than standard methods.
Transformers suffer from the computational overhead of their quadratic dependence on the length of sequences processed. We present three methods, all adding an intermediate step between training and inference/generation, which extend the autoregressive length of transformers. All rely on a Maximum Entropy Principle (MEP) whereby entropy is maximized in the presence of suitable constraints, accounted for by use of Lagrange Multipliers. These constraint methods extend the autoregressive character from T to 2T tokens in a linear-with-T fashion. There is overhead associated with this added step, but they should still be faster than the standard methods.