transformers_ner_ja

Japanese NER with Transformers + PyTorch-Lightning + MLflow Tracking

GPU Training

build: docker build -t trf-ner-ja-train .
run: docker run --rm --gpus all -v /where/to/workspace:/app/workspace trf-ner-ja-train
- NOTE: set export GPUS=1 in run_ner.sh

build: docker-compose build
run: docker-compose up
- NOTE: check ./workspace/mlruns/0/xxx is created for each runs
- NOTE: GPU support in docker-compose will be released in 1.28.0: See. docker/compose#7929
view: open http://localhost:5000/ in your browser

Name		Name	Last commit message	Last commit date
Latest commit History 12 Commits
workspace		workspace
.gitignore		.gitignore
Dockerfile		Dockerfile
Dockerfile_mlflow		Dockerfile_mlflow
LICENSE		LICENSE
README.md		README.md
doccat.py		doccat.py
docker-compose.yml		docker-compose.yml
ner.py		ner.py
requirements.txt		requirements.txt
run_doccat.sh		run_doccat.sh
run_ner.sh		run_ner.sh