- try installing llama-cpp-python and llama-cpp-python[server] from pip ... WITH ggml-metal.metal (WITH Metal GPU support) file in python executable directory
- from previous test
- llama-cpp-python[server] FAILS
#Download from https://unofficial-builds.nodejs.org/download/release/ the appropriate build for armv6l, example https://unofficial-builds.nodejs.org/download/release/v18.9.1/node-v18.9.1-linux-armv6l.tar.gz | |
wget https://unofficial-builds.nodejs.org/download/release/v18.9.1/node-v18.9.1-linux-armv6l.tar.gz | |
tar -xzf node-v18.9.1-linux-armv6l.tar.gz | |
cd node-v18.9.1-linux-armv6l | |
sudo cp -R * /usr/local | |
node -v |
export AZ_MAIN_NAME=Kube2 | |
export AZ_RG=RG_$AZ_MAIN_NAME | |
export AZ_VNET=VNET_$AZ_MAIN_NAME | |
export AZ_IP=Public_IP_$AZ_MAIN_NAME | |
export AZ_SUBNET=Subnet_$AZ_MAIN_NAME | |
export AZ_NSG=NetworkSecurityGroup_$AZ_MAIN_NAME | |
export AZ_NAME=VM_$AZ_MAIN_NAME | |
export AZ_NIC=NIC_$AZ_MAIN_NAME | |
ENV | grep AZ |
sudo systemctl stop [servicename] | |
sudo systemctl disable [servicename] | |
#rm /etc/systemd/system/[servicename] | |
#rm /etc/systemd/system/[servicename] symlinks that might be related | |
sudo systemctl daemon-reload | |
sudo systemctl reset-failed |
from time import sleep | |
import ssl | |
import json | |
import os | |
from paho.mqtt.client import Client | |
username = "your VRM email" | |
password = "your VRM pasword" | |
portal_id = "your VRM portal ID" |
{ | |
"name": "My New Config Preset", | |
"load_params": { | |
"n_ctx": 1500, | |
"n_batch": 512, | |
"rope_freq_base": 10000, | |
"rope_freq_scale": 1, | |
"n_gpu_layers": 1, | |
"use_mlock": true, | |
"main_gpu": 0, |
# BTW: we are running in a nvidia/cuda:11.x.x-devel-ubuntu22.04 | |
# install some extra Ubuntu packages | |
apt install unzip libopenblas-dev nano git-lfs aria2c jq build-essential python3 python3-pip git -y | |
pip install --upgrade pip setuptools wheel | |
# clone llama.cpp repo | |
cd /workspace | |
git clone https://github.com/ggerganov/llama.cpp.git |