In your docker file write FROM iwitaly/nlp:cpu
for CPU version or FROM iwitaly/nlp:gpu
for GPU version.
Find the image at Docker hub https://hub.docker.com/r/iwitaly/nlp/
Modern NLP frameworks including deep learning:
- torch==0.4.1
- flair==0.3.2
- spacy==2.0.11
- dateparser==0.7.0
- pymorphy2==0.8
- yargy==0.11.0
- natasha==0.10.0
- nltk==3.2.1
- yake==0.3.7
docker build -t nlp-cpu -f ./Dockerfile.cpu .
docker build -t nlp-gpu -f ./Dockerfile.gpu .
Please keep in mind that this version works with CUDA 10. Choose another base image if your GPU does not support it.
You can find NVIDIA images here https://hub.docker.com/r/nvidia/cuda/
docker run -it --runtime=nvidia iwitaly/nlp:gpu nvidia-smi
You can also pass CUDA_VISIBLE_DEVICES
environment variable.