TY  - RPRT
TI  - The Stability-Efficiency Dilemma: Investigating Sequence Length Warmup for Training GPT Models
AU  - Conglong Li
AU  - Minjia Zhang
AU  - Yuxiong He
PY  - 2022
UR  - https://arxiv.org/abs/2108.06084
ID  - 2108.06084
ER  -