Scholarlink

So let’s say you want to use PCGML to generate something, which seems a reasonable assumption if you’re reading this book. In this chapter we’ll overview a high level process for doing just that, discussing practical and ethical considerations as we go. We break this process into four steps: (1) produce or acquire training data, (2) train the model, (3) generate content from the model, and (4) evaluate the output content. We’ll first overview these steps as an outline then dive into each of them in detail. We present a visualization of this process as a reference in Fig. 4.1.

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

PCGML Process Overview

  • 章节

  • 发表日期：

    2025年

查看全文
  • Matthew Guzdial,
  • Sam Snodgrass,
  • Adam Summerville

摘要

So let’s say you want to use PCGML to generate something, which seems a reasonable assumption if you’re reading this book. In this chapter we’ll overview a high level process for doing just that, discussing practical and ethical considerations as we go. We break this process into four steps: (1) produce or acquire training data, (2) train the model, (3) generate content from the model, and (4) evaluate the output content. We’ll first overview these steps as an outline then dive into each of them in detail. We present a visualization of this process as a reference in Fig. 4.1.