Sign In

๐•ฎ๐–†๐–‹๐–Š Style (Cafe Style) [MOSS]

3

Updated: Nov 21, 2026

style

Not recently loaded

Download

1 variant available

fp16 SafeTensor

cafe_e39_s2067.safetensors

Half precision, best balance โ€ข 141.83 MB

Verified:

Type
LoRA
Stats

34

3

Reviews
Published

Nov 21, 2026

Base Model
Training
Steps: 2,067Epochs: 39
Usage Tips
Strength: 1
Hash
AutoV2001F86C030
ChatGPT ๅ›พๅƒ 2026ๅนด11ๆœˆ21ๆ—ฅ 17_51_18.png

No trigger word is needed.

Dataset was taken from NovelAI generated images in posts by Tieba user ๐•ฎ๐–†๐–‹๐–Š (ID 308119172). The training images ranged in resolution from 0.7 MP to 4 MP. The annotation methods included pure booru tags, English natural language, and JSON Layout. It was trained based on MOSS-Image-heal, and testing has shown that it can also achieve good results when used on MOSS-Image-Base.


ไธ้œ€่ฆ่งฆๅ‘่ฏ

ๆ•ฐๆฎ้›†ๅ–่‡ช่ดดๅง็”จๆˆท ๐•ฎ๐–†๐–‹๐–Š (ID 308119172) ๅธ–ๅญไธญ็š„ NovelAI ็”Ÿๆˆ็š„ๅ›พ็‰‡๏ผŒ่ฎญ็ปƒ็”จ็š„ๅ›พ็‰‡ๅˆ†่พจ็އไปŽ 0.7 MP ๅˆฐ 4MP ไธ็ญ‰๏ผŒๆ‰“ๆ ‡ๆ–นๅผๅŒ…ๆ‹ฌไบ†็บฏ booru tagใ€่‹ฑๆ–‡่‡ช็„ถ่ฏญ่จ€ใ€JSON Layout๏ผŒๅŸบไบŽ MOSS-Image-heal ่ฎญ็ปƒ๏ผŒ็ป่ฟ‡ๆต‹่ฏ•๏ผŒๅœจ MOSS-Image-Base ไธŠไฝฟ็”จไบฆๅฏ่Žทๅพ—ๆฏ”่พƒๅฅฝ็š„ๆ•ˆๆžœ