This is the MLBox Prototype. This is still under construction, some parts probably don't work yet, or may have unexpected/inconsistent behaviours.

You can get the MLBox project as a zip file for direct download from https://github.com/mlperf/mlbox/archive/master.zip

Expand this locally on the filesystem in your location of choice.

You can clone the MLBox project using Git from https://github.com/mlperf/mlbox.git

git clone https://github.com/mlperf/mlbox.git

After downloading or cloning, you can install:

python setup.py bdist_wheel sdist
pip install dist/mlbox-0.0.1-py3-none-any.whl

To uninstall:

pip uninstall mlbox

To run the toy implementation (aka "fake model"):

cd mlbox
python mlbox_run.py ../examples/fake_model:train/small_batch

--log_file=/tmp/my_log_file

To run the transformer implementation:

cd mlbox/
python mlbox_local_run.py ../examples/transformer:downloaddata/default
python mlbox_local_run.py ../examples/transformer:preprocess/default
python mlbox_local_run.py ../examples/transformer:train/default

Check out the examples directory for detailed examples.

mlperf/mlbox is licensed under the Apache License 2.0.

See https://github.com/mlperf/mlbox/blob/master/LICENSE for more information

Create an issue https://github.com/mlperf/mlbox/issues/new/choose