Configuration for TabbyAPI. See https://github.com/theroyallab/tabbyAPI/wiki/02.-Server-options and https://github.com/theroyallab/tabbyAPI/blob/main/config_sample.yml.
Declarations
Type
open submodule of (YAML 1.1 value)Example
{
model = {
model_name = "Qwen3_5-9B";
max_seq_len = 32768;
cache_mode = "8,8";
tensor_parallel = true;
gpu_split = [ 16 24 ];
};
draft_model.draft_mode = "ngram";
memory.sysmem_kv_cache = 4096;
}