Dockerfile_predict

FROM pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime

ENV PYTHONPATH=/

RUN echo ". /opt/conda/etc/profile.d/conda.sh" >> ~/.bashrc && \
    echo "conda activate base" >> ~/.bashrc

SHELL ["/bin/bash", "-c"]
WORKDIR /

# Install requirements
COPY requirements_pip.txt /
RUN pip3 install -r /requirements_pip.txt

# pre-download tokenizer and models
RUN mkdir /models
RUN mkdir /tokenizer
COPY download_models_and_tokenizer.py /
RUN python3 download_models_and_tokenizer.py

# copy script files
COPY common_training /common_training/
COPY predict.py /predict.py

# switch to local tokenizers
RUN sed -i -e 's/microsoft\/deberta-v2-xxlarge/\/tokenizer\/microsoft\/deberta-v2-xxlarge/g' predict.py
RUN sed -i -e 's/FacebookAI\/xlm-roberta-large/\/tokenizer\/FacebookAI\/xlm-roberta-large/g' predict.py

RUN echo '#!/bin/bash' > /run.sh && \
echo 'python3 /predict.py --sentences-file /dataset/sentences.tsv --output-file /output/run.tsv --models-dir /models' >> /run.sh

ENTRYPOINT [ "sh", "/run.sh" ]