Transformers Without Normalization: Dynamic Tanh Achieves Strong Performance
An episode of the Build Wiz AI Show podcast, hosted by Build Wiz AI, titled "Transformers Without Normalization: Dynamic Tanh Achieves Strong Performance" was published on March 24, 2025 and runs 11 minutes.
March 24, 2025 ·11m · Build Wiz AI Show
Summary
This podcast episode delves into the "Transformers without Normalization" paper, which introduces Dynamic Tanh (DyT) as a potential replacement for normalization layers in Transformers. DyT, a simple operation defined as tanh(αx) with a learnable parameter, aims to replicate the effects of Layer Norm without calculating activation statistics. Could DyT offer similar or better performance and improved efficiency, challenging the necessity of normalization in modern neural networks?
Episode Description
This podcast episode delves into the "Transformers without Normalization" paper, which introduces Dynamic Tanh (DyT) as a potential replacement for normalization layers in Transformers. DyT, a simple operation defined as tanh(αx) with a learnable parameter, aims to replicate the effects of Layer Norm without calculating activation statistics. Could DyT offer similar or better performance and improved efficiency, challenging the necessity of normalization in modern neural networks?
Similar Episodes
Jan 15, 2016 ·18m
Dec 23, 2015 ·40m
Dec 18, 2015 ·9m
Dec 7, 2015 ·16m
Nov 11, 2015 ·10m