conftrace_

layer norm

3 papers

Explore in graph

Co-occurring keywords · papers mentioning both

model performance (1) large language model (1) editing accuracy (1) batch norm (1) progressive replacement (1) re-parameterized batchnorm (1) slab (1) performance degradation (1) linear attention (1) language model (1)

Papers

Anisotropy is Not Inherent to Transformers NAACL 2024 Reasons and Solutions for the Decline in Model Performance after Editing NeurIPS 2024 SLAB: Efficient Transformers with Simplified Linear Attention and Progressive Re-parameterized Batch Normalization ICML 2024