/picard

PICARD - Parsing Incrementally for Constrained Auto-Regressive Decoding from Language Models

Primary LanguageHaskellApache License 2.0Apache-2.0

= to fine-tune model =

  • place fine-tune data in seq2seq/database/insurance/insurance.json (example included in repo)
  • run "make train" to open a docker container
  • in separate terminal run ./custom_train.sh

= to run predictions=