A Docker image with an installed perceiver-io
library is available on the GitHub Container registry. Training runs can be started with:
sudo docker run \
-v $(pwd)/.cache:/app/.cache \
-v $(pwd)/logs:/app/logs \
--rm \
--ipc=host \
--name=perceiver-io \
--runtime=nvidia \
ghcr.io/krasserm/perceiver-io:latest \
python -m SCRIPT fit [OPTIONS]
where SCRIPT
must be replaced by the module name of a training script and [OPTIONS]
with the training script
options. For example:
sudo docker run \
-v $(pwd)/.cache:/app/.cache \
-v $(pwd)/logs:/app/logs \
--rm \
--ipc=host \
--name=perceiver-io \
--runtime=nvidia \
ghcr.io/krasserm/perceiver-io:latest \
python -m perceiver.scripts.text.mlm fit \
--model.params=deepmind/language-perceiver \
...