Memorization in Attention-only Transformers
This work addresses a foundational issue in language modeling by providing more accurate memorization bounds, which is incremental but important for model analysis.
The paper tackles the problem of understanding memorization capacity in attention-only Transformers by extending prior hypotheses to any context size, achieving more effective exact memorization and introducing approximate memorization of distributions, with experimental validation showing improved bounds.
Recent research has explored the memorization capacity of multi-head attention, but these findings are constrained by unrealistic limitations on the context size. We present a novel proof for language-based Transformers that extends the current hypothesis to any context size. Our approach improves upon the state-of-the-art by achieving more effective exact memorization with an attention layer, while also introducing the concept of approximate memorization of distributions. Through experimental validation, we demonstrate that our proposed bounds more accurately reflect the true memorization capacity of language models, and provide a precise comparison with prior work.