Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-26 12:21:40 +01:00
Code Issues Packages Projects Releases Wiki Activity

scripts : add non-interactive server-llm.sh (#5303)

Browse Source 
* Update server-llm.sh

Add flag --non-interactive that allows run script without asking a permission

* Update scripts/server-llm.sh

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Нияз Гарифзянов 2024-02-05 10:43:57 +03:00 committed by GitHub
parent 5d55b0cd82
commit 4be04c8965
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 40 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
scripts/server-llm.sh
View File

@ -47,6 +47,7 @@ if ! command -v make &> /dev/null; then
fi
# parse arguments
is_interactive=1
port=8888
repo=""
wtype=""
@ -66,15 +67,16 @@ verbose=0
function print_usage {
printf "Usage:\n"
printf " ./server-llm.sh [--port] [--repo] [--wtype] [--backend] [--gpu-id] [--n-parallel] [--n-kv] [--verbose]\n\n"
printf " --port: port number, default is 8888\n"
printf " --repo: path to a repo containing GGUF model files\n"
printf " --wtype: weights type (f16, q8_0, q4_0, q4_1), default is user-input\n"
printf " --backend: cpu, cuda, metal, opencl, depends on the OS\n"
printf " --gpu-id: gpu id, default is 0\n"
printf " --n-parallel: number of parallel requests, default is 8\n"
printf " --n-kv: KV cache size, default is 4096\n"
printf " --verbose: verbose output\n\n"
printf " ./server-llm.sh [-interactive] [--port] [--repo] [--wtype] [--backend] [--gpu-id] [--n-parallel] [--n-kv] [--verbose]\n\n"
printf " --non-interactive: run without asking a permision to run\n"
printf " --port: port number, default is 8888\n"
printf " --repo: path to a repo containing GGUF model files\n"
printf " --wtype: weights type (f16, q8_0, q4_0, q4_1), default is user-input\n"
printf " --backend: cpu, cuda, metal, opencl, depends on the OS\n"
printf " --gpu-id: gpu id, default is 0\n"
printf " --n-parallel: number of parallel requests, default is 8\n"
printf " --n-kv: KV cache size, default is 4096\n"
printf " --verbose: verbose output\n\n"
printf "Example:\n\n"
printf ' bash -c "$(curl -s https://ggml.ai/server-llm.sh)"\n\n'
}
@ -82,6 +84,10 @@ function print_usage {
while [[ $# -gt 0 ]]; do
key="$1"
case $key in
--non-interactive)
is_interactive=0
shift
;;
--port)
port="$2"
shift
@ -176,31 +182,32 @@ repos=(
"https://huggingface.co/TheBloke/OpenHermes-2-Mistral-7B-GGUF"
"https://huggingface.co/TheBloke/CausalLM-7B-GGUF"
)
if [ $is_interactive -eq 1 ]; then
printf "\n"
printf "[I] This is a helper script for deploying llama.cpp's server on this machine.\n\n"
printf " Based on the options that follow, the script might download a model file\n"
printf " from the internet, which can be a few GBs in size. The script will also\n"
printf " build the latest llama.cpp source code from GitHub, which can be unstable.\n"
printf "\n"
printf " Upon success, an HTTP server will be started and it will serve the selected\n"
printf " model using llama.cpp for demonstration purposes.\n"
printf "\n"
printf " Please note:\n"
printf "\n"
printf " - All new data will be stored in the current folder\n"
printf " - The server will be listening on all network interfaces\n"
printf " - The server will run with default settings which are not always optimal\n"
printf " - Do not judge the quality of a model based on the results from this script\n"
printf " - Do not use this script to benchmark llama.cpp\n"
printf " - Do not use this script in production\n"
printf " - This script is only for demonstration purposes\n"
printf "\n"
printf " If you don't know what you are doing, please press Ctrl-C to abort now\n"
printf "\n"
printf " Press Enter to continue ...\n\n"
printf "\n"
printf "[I] This is a helper script for deploying llama.cpp's server on this machine.\n\n"
printf " Based on the options that follow, the script might download a model file\n"
printf " from the internet, which can be a few GBs in size. The script will also\n"
printf " build the latest llama.cpp source code from GitHub, which can be unstable.\n"
printf "\n"
printf " Upon success, an HTTP server will be started and it will serve the selected\n"
printf " model using llama.cpp for demonstration purposes.\n"
printf "\n"
printf " Please note:\n"
printf "\n"
printf " - All new data will be stored in the current folder\n"
printf " - The server will be listening on all network interfaces\n"
printf " - The server will run with default settings which are not always optimal\n"
printf " - Do not judge the quality of a model based on the results from this script\n"
printf " - Do not use this script to benchmark llama.cpp\n"
printf " - Do not use this script in production\n"
printf " - This script is only for demonstration purposes\n"
printf "\n"
printf " If you don't know what you are doing, please press Ctrl-C to abort now\n"
printf "\n"
printf " Press Enter to continue ...\n\n"
read
read
fi
if [[ -z "$repo" ]]; then
printf "[+] No repo provided from the command line\n"