My SillyTavern settings
Hello dude, I really appreciate the effort, but can I ask you to upload it on https://catbox.moe/ ?
I can do it if you want, because for whatever reason there is a limit of download on your file.
Context : https://files.catbox.moe/vnnsra.json
Instruct : https://files.catbox.moe/7sudj5.json
Preset : https://files.catbox.moe/og94e5.json
Sorry, but I see the download count get down, and I want to save this. Thanks again!
Context : https://files.catbox.moe/vnnsra.json
Instruct : https://files.catbox.moe/7sudj5.json
Preset : https://files.catbox.moe/og94e5.jsonSorry, but I see the download count get down, and I want to save this. Thanks again!
You did the right thing man. Thank you ;)
Sorry but for you this may be very trivial, but for beginners causes questions:
Quote from model card:
"You need to use our custom prompting format(scroll down to see them! or just directly download the SillyTavern instruct preset here)"
Could you please explain where we have to put the json file in Sillytavern? (add it via gui (How?Where? or put it somewhere in the folderstructure of SillyTavern?)
Do we need those files as well? Or are they optional (for experimenting and feedback)?
"
Context : https://files.catbox.moe/vnnsra.json
Instruct : https://files.catbox.moe/7sudj5.json
Preset : https://files.catbox.moe/og94e5.json
"
Sorry but for you this may be very trivial, but for beginners causes questions:
Quote from model card:
"You need to use our custom prompting format(scroll down to see them! or just directly download the SillyTavern instruct preset here)"Could you please explain where we have to put the json file in Sillytavern? (add it via gui (How?Where? or put it somewhere in the folderstructure of SillyTavern?)
Do we need those files as well? Or are they optional (for experimenting and feedback)?
"
Context : https://files.catbox.moe/vnnsra.json
Instruct : https://files.catbox.moe/7sudj5.json
Preset : https://files.catbox.moe/og94e5.json
"
In each of the setting, you have this button:
Then you can import a json file.
We provided only the Instruct json, that's all you need to run it basically.
The OP provided Advanced Formatting, Instruct, and Preset settings, you can import them too in the same way.
Trying the model out on OpenRouter now that it's on there. What selection for tokenizer? Mistral?
Can someone help? I'm running this model using llama.cpp with the following command
./server -m $MODEL -c 32768
With SillyTavern using the setting json files in this post, but the model keeps endlessly generating, sometimes with some special sequences in the text, such as square brackets [ ] or <im_end> <im_start>
I have never understood how to configure Mixtral with llama.cpp and SillyTavern properly, I've tried 3 Mixtral models with and without the default Mixtral configuration on SillyTavern and the result is the same :(