hrsd-vqa

Build image

docker build -t "ns-hrsd-image" .

setup docker

sudo apt update sudo apt install apt-transport-https ca-certificates curl software-properties-common -y

curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg echo "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list sudo apt update sudo apt install docker-ce docker-ce-cli containerd.io -y sudo usermod -aG docker ${USER} su - ${USER} sudo systemctl enable docker

Setup docker GPU

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
sudo apt-get update && sudo apt-get install -y nvidia-docker2
sudo systemctl restart docker

Run training

docker run -it --rm   -v /home/ubuntu/pix2struct-vqa-dummy-data/edc_dummy_data:/usr/src/data   -e MAX_PATCHES=3072   -e MAX_LENGTH=256 -e NUM_EPOCHS=5   --gpus all --ipc=host  neuralspaceacr.azurecr.io/hrsd/ns-hrsd-qna:v1   python dist/finetune.py

Run inference

sudo docker run -p 8003:8003 --gpus all -v /home/elias/neuralspace/dataset/edc_dummy_data/training/model:/usr/src/model -e MODEL_PATH=/usr/src/model neuralspaceacr.azurecr.io/hrsd/ns-hrsd-qna:v1 uvicorn server:app --host 0.0.0.0 --port 8003

Set the model path to the trained model folder. Once training is finished, it can be found inside training folder in the data folder. For using the default model, remove the MODEL_PATH environment variable from above command.

docker run -it --rm -v /home/elias/neuralspace/dataset/edc_dummy_data:/usr/src/data -e MAX_PATCHES=3072 -e NUM_GPUS=2 -e BATCH_SIZE=2 -e MAX_LENGTH=256 -e NUM_EPOCHS=5 --gpus all --ipc=host neuralspaceacr.azurecr.io/hrsd/ns-hrsd-qna:v1 python dist/finetune.py

sudo docker run -p 8003:8003 --gpus all -v /home/elias/neuralspace/dataset/edc_dummy_data/training/model:/usr/src/model -e MODEL_PATH=/usr/src/model neuralspaceacr.azurecr.io/hrsd/ns-hrsd-qna:v1 uvicorn server:app --host 0.0.0.0 --port 8003