The repository contains code to reproduce models from the paper Adapting LLMs for Minimal-edit Grammatical Error Correction.
The datasets can be downloaded from here.
Before running the script install environment from env.yml file.
For inference we recommend using the vLLM framework (code available in inf_model.py file).