Multi-Tagger v1.4

Update:
1. Decided to remove Prompt Enhancer and 'Beautify' because there's already several versions of them out there and I want to use less troubling outdated imports to speed up start up.
2. Added 'explicitly unload the current model from memory' to avoid models getting stuck on VRAM when you are not using it, so it's also useful if you are using multiple AI tools.
3. Added PixAI and is 100% working, but it can be a little or very slow to use on CPU, so I recommend cloning this repo for local use or try it on a GPU powered space, it takes only 3s or less to tag any image if you use a GPU, it's so fast!
4. Finally fixed the gallery, so it always returns tags from the first image even if you are tagging multiple images and it was more easy than I thought...

Anyway, have fun!

app.py

@@ -24,7 +24,7 @@ pip install ...
 python app.py
 """
 
-TITLE = 'Multi-Tagger v1.3'
+TITLE = 'Multi-Tagger v1.4'
 DESCRIPTION = '\nMulti-Tagger is a versatile application for advanced image analysis and captioning. Supports <b>CUDA</b> and <b>CPU</b>.\n'
 
 SWINV2_MODEL_DSV3_REPO = 'SmilingWolf/wd-swinv2-tagger-v3'