Somebody know how to have this parameters on koboldcpp? sorry for the basic question but i cant get the same performance than using llama.cpp when using kaboldcpp :(

[kroryan]

Hello so i realize if i run the LLM wit this parameters on llama.cpp it goes faster and give good answer so i want to have the same parameter with koboldcpp because when i use it on koboldcpp it is superslow here the parameters:

./main -m ./models/ggml-vic7b-uncensored-q5_0.bin -c 512 -b 1024 -n 256 --keep 48
--repeat_penalty 1.0 --color -i
-r "User:" -f prompts/chat-with-bob.txt

I know about the prompt, it is not really important, Just the others parameters, im using a smartphone to run it and as i said i get much more better resulta using llama.cpp with this command than using koboldcpp

Note: i think the ones are making it Work well are: -n 256 --keep 48