Hacker News:AI 热帖· qtqtqt·· 2 小时前AI 评分72
Amazon Web Services 论文提出 Priming:无需从头训练即可改造 Transformer 架构
Can we change an AI model's architecture without retraining it from scratch?
AI 导读
Amazon Web Services 团队在论文 Priming: Hybrid state space models from pre-trained Transformers 中提出 priming 方法,用预训练 Transformer 权重初始化混合状态空间模型,并将预训练成本降低超过 99%。
来源:Hacker News:AI 热帖 · amazon.science