BertInterlinear — BERT for Interlinear Analysis

BertInterlinear applies BERT (Bidirectional Encoder Representations from Transformers) to interlinear analysis tasks. It includes multiple Python scripts demonstrating various BERT applications such as active fill-in-the-blank, masked language modeling, and next-word prediction — all fully Dockerized for easy deployment.

Usage

git clone https://github.com/ebadi/fitbert.git

docker build -f Dockerfile -t bertinterlinear:latest .
docker run -v=$PWD:/data/ bertinterlinear:latest /bin/bash -c "python3 /data/activebert.py"
docker run -v=$PWD:/data/ bertinterlinear:latest /bin/bash -c "python3 /data/fitbertx.py"
docker run -v=$PWD:/data/ bertinterlinear:latest /bin/bash -c "python3 /data/betterfitbert.py"

The Docker image installs the BERT models during build, so no separate download step is needed.

Links

BertInterlinear GitHub repository