Generalized Unique Hard Attention

Back to List
method ID: generalizeduniquehardattention

Related Papers

Formal Language Recognition by Hard Attention Transformers: Perspectives from Circuit Complexity

Authors: nan

Publication: TIPSTER (2022)

DOI: nan

AbstractThis paper analyzes three formal models of Transformer encoders that differ in the form of their self-attention mechanism: unique hard attention (UHAT); generalized unique hard attention (GUHAT), which generalizes UHAT; …...

26740