Built based on https://sparknlp.org/docs/en/install#docker-support), updated to later sparknlp and pyspark versions. Because of conflict dependency (six) used by tensorflow 2.12.0 and scikit-spark, remove tensorflow from this image (spark people says it's only used for Visual NLP). Current build uses: pyspark==3.5.0, spark-nlp==5.2.0. Based on my experiment, sparknlp doesn't work with torch. So no torch for this image either.
The built image is published here: https://hub.docker.com/r/brokenjade/sparknlp_gpu
Content type
Image
Digest
sha256:f92594ac8…
Size
13.6 GB
Last updated
over 2 years ago
docker pull brokenjade/sparknlp_gpu