mirror of
https://github.com/fauxpilot/fauxpilot.git
synced 2025-08-20 13:24:27 -07:00
Simplify config and port handling
This commit is contained in:
parent
6f49915d2a
commit
87f4f53e27
5 changed files with 70 additions and 45 deletions
93
setup.sh
93
setup.sh
|
@ -1,9 +1,15 @@
|
|||
#!/bin/bash
|
||||
|
||||
if [ -f config.env ]; then
|
||||
echo "config.env already exists, skipping"
|
||||
echo "Please delete config.env if you want to re-run this script"
|
||||
exit 0
|
||||
if [ -f .env ]; then
|
||||
read -rp ".env already exists, do you want to delete .env and recreate it? [y/n] " DELETE
|
||||
if [[ ${DELETE:-y} =~ ^[Yy]$ ]]
|
||||
then
|
||||
echo "Deleting .env"
|
||||
rm .env
|
||||
else
|
||||
echo "Exiting"
|
||||
exit 0
|
||||
fi;
|
||||
fi
|
||||
|
||||
echo "Models available:"
|
||||
|
@ -16,7 +22,7 @@ echo "[6] codegen-6B-multi (13GB total VRAM required; multi-language)"
|
|||
echo "[7] codegen-16B-mono (32GB total VRAM required; Python-only)"
|
||||
echo "[8] codegen-16B-multi (32GB total VRAM required; multi-language)"
|
||||
# Read their choice
|
||||
read -p "Enter your choice [6]: " MODEL_NUM
|
||||
read -rp "Enter your choice [6]: " MODEL_NUM
|
||||
|
||||
# Convert model number to model name
|
||||
case $MODEL_NUM in
|
||||
|
@ -32,45 +38,68 @@ case $MODEL_NUM in
|
|||
esac
|
||||
|
||||
# Read number of GPUs
|
||||
read -p "Enter number of GPUs [1]: " NUM_GPUS
|
||||
read -rp "Enter number of GPUs [1]: " NUM_GPUS
|
||||
NUM_GPUS=${NUM_GPUS:-1}
|
||||
|
||||
read -rp "External port for the API [5000]: " API_EXTERNAL_PORT
|
||||
API_EXTERNAL_PORT=${API_EXTERNAL_PORT:-5000}
|
||||
|
||||
read -rp "Address for Triton [triton]: " TRITON_HOST
|
||||
TRITON_HOST=${TRITON_HOST:-triton}
|
||||
|
||||
read -rp "Port of Triton host [8001]: " TRITON_PORT
|
||||
TRITON_PORT=${TRITON_PORT:-8001}
|
||||
|
||||
# Read model directory
|
||||
read -p "Where do you want to save the model [$(pwd)/models]? " MODEL_DIR
|
||||
read -rp "Where do you want to save the model [$(pwd)/models]? " MODEL_DIR
|
||||
if [ -z "$MODEL_DIR" ]; then
|
||||
MODEL_DIR="$(pwd)/models"
|
||||
else
|
||||
MODEL_DIR="$(readlink -m "${MODEL_DIR}")"
|
||||
fi
|
||||
|
||||
# Write config.env
|
||||
echo "MODEL=${MODEL}" > config.env
|
||||
echo "NUM_GPUS=${NUM_GPUS}" >> config.env
|
||||
echo "MODEL_DIR=${MODEL_DIR}" >> config.env
|
||||
# Write .env
|
||||
echo "MODEL=${MODEL}" > .env
|
||||
echo "NUM_GPUS=${NUM_GPUS}" >> .env
|
||||
echo "MODEL_DIR=${MODEL_DIR}/${MODEL}-${NUM_GPUS}gpu" >> .env
|
||||
echo "API_EXTERNAL_PORT=${API_EXTERNAL_PORT}" >> .env
|
||||
echo "TRITON_HOST=${TRITON_HOST}" >> .env
|
||||
echo "TRITON_PORT=${TRITON_PORT}" >> .env
|
||||
GPUS="$(seq -s ',' -t '\n' "${NUM_GPUS}")"
|
||||
echo "GPUS=${GPUS%?}" >> .env
|
||||
|
||||
if [ -d "$MODEL_DIR"/"${MODEL}"-"${NUM_GPUS}"gpu ]; then
|
||||
echo "Converted model for ${MODEL}-${NUM_GPUS}gpu already exists."
|
||||
read -p "Do you want to re-use it? y/n: " REUSE_CHOICE
|
||||
if ${REUSE_CHOICE^^} != "Y"; then
|
||||
# Create model directory
|
||||
mkdir -p "${MODEL_DIR}"
|
||||
|
||||
# For some of the models we can download it pre-converted.
|
||||
if [ "$NUM_GPUS" -le 2 ]; then
|
||||
echo "Downloading the model from HuggingFace, this will take a while..."
|
||||
SCRIPT_DIR="$(dirname "$(readlink -f "$0")")"
|
||||
DEST="${MODEL}-${NUM_GPUS}gpu"
|
||||
ARCHIVE="${MODEL_DIR}/${DEST}.tar.zst"
|
||||
cp -r "$SCRIPT_DIR"/converter/models/"$DEST" "${MODEL_DIR}"
|
||||
curl -L "https://huggingface.co/moyix/${MODEL}-gptj/resolve/main/${MODEL}-${NUM_GPUS}gpu.tar.zst" \
|
||||
-o "$ARCHIVE"
|
||||
zstd -dc "$ARCHIVE" | tar -xf - -C "${MODEL_DIR}"
|
||||
rm -f "$ARCHIVE"
|
||||
else
|
||||
echo "Downloading and converting the model, this will take a while..."
|
||||
docker run --rm -v "${MODEL_DIR}":/models -e MODEL="${MODEL}" -e NUM_GPUS="${NUM_GPUS}" moyix/model_converter:latest
|
||||
fi
|
||||
read -rp "Do you want to re-use it? y/n: " REUSE_CHOICE
|
||||
if [ "${REUSE_CHOICE^^}" = "Y" ]; then
|
||||
exit 0
|
||||
fi
|
||||
fi
|
||||
|
||||
echo "Done! Now run ./launch.sh to start the FauxPilot server."
|
||||
# Create model directory
|
||||
mkdir -p "${MODEL_DIR}"
|
||||
|
||||
# For some of the models we can download it pre-converted.
|
||||
if [ $NUM_GPUS -le 2 ]; then
|
||||
echo "Downloading the model from HuggingFace, this will take a while..."
|
||||
SCRIPT_DIR="$(dirname "$(readlink -f "$0")")"
|
||||
DEST="${MODEL}-${NUM_GPUS}gpu"
|
||||
ARCHIVE="${MODEL_DIR}/${DEST}.tar.zst"
|
||||
cp -r "$SCRIPT_DIR"/converter/models/"$DEST" "${MODEL_DIR}"
|
||||
curl -L "https://huggingface.co/moyix/${MODEL}-gptj/resolve/main/${MODEL}-${NUM_GPUS}gpu.tar.zst" \
|
||||
-o "$ARCHIVE"
|
||||
zstd -dc "$ARCHIVE" | tar -xf - -C "${MODEL_DIR}"
|
||||
rm -f "$ARCHIVE"
|
||||
else
|
||||
echo "Downloading and converting the model, this will take a while..."
|
||||
docker run --rm -v "${MODEL_DIR}":/models -e MODEL=${MODEL} -e NUM_GPUS="${NUM_GPUS}" moyix/model_converter:latest
|
||||
fi
|
||||
|
||||
read -rp "Config complete, do you want to run FauxPilot? [y/n]" RUN
|
||||
if [[ ${RUN:-y} =~ ^[Yy]$ ]]
|
||||
then
|
||||
bash ./launch.sh
|
||||
else
|
||||
echo "You can run ./launch.sh to start the FauxPilot server."
|
||||
exit 0
|
||||
fi;
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue