update

Dockerfile.ARM

@@ -0,0 +1,34 @@
+FROM nvcr.io/nvidia/cuda-arm64:11.4.0-devel-ubuntu20.04
+LABEL maintainer="bigscience-workshop"
+LABEL repository="petals"
+
+WORKDIR /home
+# Set en_US.UTF-8 locale by default
+RUN echo "LC_ALL=en_US.UTF-8" >> /etc/environment
+
+# Install packages
+RUN apt-get update && apt-get install -y --no-install-recommends \
+  build-essential \
+  wget \
+  git \
+  sed \
+  && apt-get clean autoclean && rm -rf /var/lib/apt/lists/{apt,dpkg,cache,log} /tmp/* /var/tmp/*
+
+RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O install_miniconda.sh && \
+  bash install_miniconda.sh -b -p /opt/conda && rm install_miniconda.sh
+ENV PATH="/opt/conda/bin:${PATH}"
+
+RUN conda install python~=3.10.12 pip && \
+    pip install --no-cache-dir "torch>=1.12" && \
+    conda clean --all && rm -rf ~/.cache/pip
+
+VOLUME /cache
+ENV PETALS_CACHE=/cache
+
+RUN sed '46d' setup.cfg
+
+COPY . petals/
+RUN pip install --no-cache-dir -e petals
+
+WORKDIR /home/petals/
+CMD bash

README.md

@@ -144,3 +144,25 @@ _arXiv preprint arXiv:2209.01188,_ 2022.
 <p align="center">
     <img src="https://petals.dev/bigscience.png" width="150">
 </p>
+
+
+# setup
+
+
+ 1623  sudo cp petals-inference.service /etc/systemd/system/
+ 1634  sudo systemctl daemon-reload
+ 1635  sudo systemctl status petals-inference.service -l
+ 1636  sudo systemctl restart petals-inference.service -l
+
+ 1639  sudo useradd petals
+ 1640  sudo mkdir /home/petals
+ 1641  sudo chown petals: /home/petals/
+ 1643  sudo cp -r ~/.venv/ /home/petals/venv
+ 1644  sudo rm -rf /home/petals/venv
+ 1658  sudo mv ~/.venv/ /home/petals/venv
+ 1659  sudo chown petals: /home/petals/
+
+1670  sudo systemctl status petals-inference.service -l
+ 1674  sudo systemctl restart petals-inference.service -l
+ 1675  sudo systemctl status petals-inference.service -l
+ 

etc/petals-inference.service

@@ -0,0 +1,11 @@
+[Unit]
+Description=Petals Inference
+
+[Service]
+#User=petals
+#Group=petals
+Environment=PJRT_DEVICE=TPU
+ExecStart=/home/petals/venv/bin/python -m petals.cli.run_server --port 31330 petals-team/StableBeluga2 --device xla --num_blocks=4
+
+[Install]
+WantedBy=multi-user.target

run.sh

@@ -1 +1,2 @@
-PJRT_DEVICE=TPU python -m petals.cli.run_server --port 31330 petals-team/StableBeluga2 --device xla --num_blocks=4  --initial_peers /ip4/10.128.0.41/tcp/8099/p2p/QmVA1BeK2UYaXPwmHQ5EZ2biyzyVWdnQj2zHcEVYxgiSDY
+PJRT_DEVICE=TPU python -m petals.cli.run_server --port 31330 petals-team/StableBeluga2 --device xla --num_blocks=4
+