fix: metadata dataset degradation and make it work

[kohya-ss]

The documentation also needs to be updated.

[Copilot]

Pull Request Overview

This PR fixes metadata dataset degradation issues and enhances the WD14 tagger functionality to support both JSON and JSONL metadata formats. The changes improve the handling of image metadata, add support for new tag categories, and provide better image preprocessing.

• Fix null comparison issues in metadata handling to prevent dataset degradation
• Add comprehensive JSONL metadata format support alongside existing JSON format
• Extend WD14 tagger with new tag categories (copyright, meta, model, quality) and improved model support

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
library/train_util.py	Fixes null checks in metadata comparison and refactors dataset initialization to better handle image paths and cache files
finetune/tag_images_by_wd14_tagger.py	Major enhancement adding JSONL support, new tag categories, improved image preprocessing, and better model handling
docs/wd14_tagger_README-ja.md	Updates Japanese documentation with current dependency versions and help command reference
docs/wd14_tagger_README-en.md	Updates English documentation with current dependency versions and help command reference

Comments suppressed due to low confidence (1)

finetune/tag_images_by_wd14_tagger.py:1

• JSOL should be JSONL (JSON Lines format).

import argparse

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}