conftrace_

loss change allocation

2 papers

Explore in graph

Also known as

LCA

Co-occurring keywords · papers mentioning both

parameter contribution (1) path integral (1) parameter analysis (1) semantic distance (1) masked language modeling (1) causal language modeling (1) training dynamics (1) machine translation (1) knowledge distillation (1) optimal transport (1)

Papers

A Closer Look at Parameter Contributions When Training Neural Language and Translation Models COLING 2022 LCA: Loss Change Allocation for Neural Network Training NeurIPS 2019