celine-lee / allennlp-srl-verb Goto Github PK

To set up the environment and run the model, follow the following workflow:

Create the virtual environment and set up installs:

python3 -m venv verb_srl_venv
cd verb_srl_venv
source bin/activate
pip install allennlp==1.0.0rc3
pip install allennlp-models==1.0.0rc3

Check that the lib/python3.6/site-packages/allennlp_models/syntax/srl/ folder has srl-eval.pl. If it does not, wget it from here and put it in that folder.

The GPUs on the CCG machines are CUDA version 10.1, so we set Pytorch back to version 1.4:

pip install torch===1.4.0 torchvision===0.5.0 -f https://download.pytorch.org/whl/torch_stable.html

Then we set up the paths and model files, which are at bert_base_srl.jsonnet and set_paths.sh. Note that these paths are set up as of 5/18/20 and only work on the CCG machines, and the paths may have changed since then.

To train:

. ./set_paths.sh
allennlp train bert_base_srl.jsonnet -s srl-bert-test --include-package allennlp_models

To evaluate:

allennlp evaluate srl-bert-test/model.tar.gz /shared/celinel/LDC2013T19/conll-formatted-ontonotes-5.0/conll-formatted-ontonotes-5.0/data/test

To predict:

allennlp predict srl-bert-model/model.tar.gz input.txt --output-file output.txt

The input and output files for the predict call are in JSON format. Inputs should be formatted such that sentences are separated into individual sentences, rather than one large paragraph.

To predict a single sentence, without an input JSON file. The output will be in TextAnnotation view format:

python predict_sentence.py srl-bert-test/model.tar.gz "The fox jumped over the fence, and the fence fell down." --output_file test_nofile_output.txt