Scholarlink

While the core of DNN training involves the automated learning of model parameters, there are several parameters that cannot be numerically optimized. These are known as hyperparameters. In this chapter, we discuss the various hyperparameters that deep learning practitioners are responsible for setting, and provide general guidance on how to best configure them.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hyperparameter Tuning

  • 章节

  • 发表日期：

    2026年

查看全文
  • Yiran Chen,
  • Hai Li,
  • Huanrui Yang

摘要

While the core of DNN training involves the automated learning of model parameters, there are several parameters that cannot be numerically optimized. These are known as hyperparameters. In this chapter, we discuss the various hyperparameters that deep learning practitioners are responsible for setting, and provide general guidance on how to best configure them.