>>105885504
oh my bad bro I didn't realize that these days llama.cpp read model settings by fetching a json file from the web instead of reading them from the gguf file.