8,435 2 months ago

To help you with prompts in image and video generation. Same as used on https://goonsai.com Support: r/goonsai/

2 months ago

cf62f31c7147 · 6.2GB ·

qwen2
·
3.09B
·
F16
{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user
Generate CivitAI-style prompts using comma-separated tags with spaces (not underscores). Format: qua
{ "num_predict": 256, "presence_penalty": 0.5, "repeat_last_n": 64, "repeat_penalty"

Readme

Overview

Trained on 100K prompt examples for image and video generation. This is a completely uncensored model, it will refuse nothing. Use at your own risk.

This is the same prompt helper that is used by goonsai.com telegram bots. You can use it directly there. @goonspromptbot

Help

This is a Completion style model. Meaning you need provide a few words of what you need and the model will try to complete with as much information as possible. Depending on the tool you use, you could improve it with a template though not required.

example

a woman dancing in swimsuit

then you will get the rest of the prompt.

quality, high resolution, elegant pose, shimmering lighting, beach setting, tropical breeze, smooth motion, round belly, long legs, head tilted slightly to one side, eyes closed gently, smile on face, water droplets on skin

You can roll the dice on output or you can follow up to refine it. Don’t be worried about strange formatting as long as the keywords are there. The models dont care about grammer, its about tokens.

You can always edit the tempalte for this Modefile and you can join the discord Discord Channel

Limitations: There are situations when the model will spit out trigger w0rds and LoRA names of what could be deleted LoRAs.