metadata

language:
  - en
library_name: allennlp
pipeline_tag: token-classification

XLM-RoBERTa Base CoBaLD parser

This version of parser is based on XLM-R Base encoder. The encoder weights were initialized from xlm-roberta-base model.

Usage

Clone https://github.com/CobaldAnnotation/CobaldParser, go to the repo directory and install requirements.
Download the model.tar.gz file. You can do it manually of via git clone https://huggingface.co/CoBaLD/xlm-roberta-base-cobald-parser.
Run allennlp predict with the following arguments on the conllu file you want to annotate with CoBaLD:

allennlp predict <PATH_TO_model.tar.gz> <INPUT_CONLLU_FILE> \
--output-file <OUTPUT_CONLLU_FILE> \
--include-package src \
--predictor morpho_syntax_semantic_predictor \
--use-dataset-reader

Now see the results in output conllu file you specified.