v0.1.3

You can now deploy Lilac to a HuggingFace dataset with just a few lines of Python, or from the CLI:

Deploy a single configuration object, and have it load entirely on the space:

ll.deploy_config(
  hf_space='nsthorat-lilac/nikhil-demo',
  # Create the space if it doesn't exist.
  create_space=True,
  config=ll.Config(datasets=[
    ll.DatasetConfig(
      namespace='local',
      name='glue_ax',
      source=ll.HuggingFaceSource(dataset_name='glue', config_name='ax'))
  ]))

Deploy a Lilac project you've loaded locally:

ll.deploy_project(
  hf_space='nsthorat-lilac/nikhil-project-demo',
  project_dir='./data',
  datasets=['local/glue_ax'],  # This is optional. If not defined, uploads all datasets.
  # Create the space if it doesn't exist.
  create_space=True)

Or via the CLI:

lilac deploy-project --project_dir='~/my_project'

Features

• Completely overhaul the deploy scripts. by @nsthorat in #725

Other Changes

• Improve the schema view to be less overwhelming. by @nsthorat in #726

Demo

• Add the textbook quality programming dataset to the demo. Clean up old datasets. by @nsthorat in #723

Full Changelog: v0.1.2...v0.1.3