Gradient norm blow-up at the end of LLM training is caused by an interaction between weight decay, normalization layers, and the learning rate schedule, and a corrected weight decay term removes it.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Why Gradients Rapidly Increase Near the End of Training
Gradient norm blow-up at the end of LLM training is caused by an interaction between weight decay, normalization layers, and the learning rate schedule, and a corrected weight decay term removes it.