More models

[medledan]

Where do you go on huggingFace to get more token models to download? Not sure where to go. Thanks for your help.

[imrayya]

Well there no real category for it on hugginFace to get models that are specific to prompt generation, but if you search under text-generation and you should find a couple

Personally, I have theses ones setup to try:

• Gustavosta/MagicPrompt-Stable-Diffusion
• Gustavosta/MagicPrompt-Dalle
• FredZhang7/distilgpt2-stable-diffusion-v2
• FredZhang7/anime-anything-promptgen-v2
• AUTOMATIC/promptgen-lexart
• AUTOMATIC/promptgen-majinai-safe
• succinctly/text2image-prompt-generator
• microsoft/Promptist

They should mention which tokenizer they've used in the model cards.

Hope that helps

[Delteuz]

Which of these models do you recommend using? and under text-generation models at hugging face, are all of them usable? How do I handpick which one of them is good?

[imrayya]

The ones I use are:

• FredZhang7/distilgpt2-stable-diffusion-v2
• Gustavosta/MagicPrompt-Dalle
• FredZhang7/anime-anything-promptgen-v2
• FredZhang7/danbooru-tag-generator

Others that I found are (haven't tested the quality):

• microsoft/Promptist
• Gustavosta/MagicPrompt-Stable-Diffusion
• succinctly/text2image-prompt-generator
• AUTOMATIC/promptgen-lexart
• AUTOMATIC/promptgen-majinai-unsafe
• AUTOMATIC/promptgen-majinai-safe
• daspartho/prompt-extend

I haven't tested it yet but theoretically most models should work assuming you have enough ram. Including the big ones but I suggest looking for ones that uses GPT2 or DistilGPT2.

Unfortunately, HuggingFace doesn't have the best organization to find stuff.

Personally, I mostly use FredZhang7/distilgpt2-stable-diffusion-v2 and FredZhang7/danbooru-tag-generator

[malo67]

I'm having some problems with adding models, every time I try to add a model to the json the extension tab for the prompt generator disappears in automatic1111. I'm thinking I'm just putting the info in wrong maybe?

I'm getting this error.

[imrayya]

I mean I can't tell without seeing the Json file but from the error, I'm assuming you're missing a comma.
Using my above example to demonstrate it

[
    {
        "Title":"Gustavosta" <-----Missing commas between the parameters
        "Tokenizer":"gpt2",
        "Model":"Gustavosta/MagicPrompt-Dalle"
    },
    {
        "Title":"FredZhang7",
        "Tokenizer":"distilgpt2",
        "Model":"FredZhang7/distilgpt2-stable-diffusion-v2"
    } <------Missing commas between the models
    {
        "Title":"Microsfot Promptist",
        "Tokenizer":"gpt2",
        "Model":"microsoft/Promptist"
    }
]

If you can't find the problem, feel free to post the JSON for me to see

[malo67]

I was just missing the comma, thank you!