Build Docker image

Hello,

I’m looking to build the deeppavlov/deeppavlov Docker image, so that:

  1. I can get the latest version 1.7.0 (the current released docker image isn’t up-to-date with 1.7.0)

  2. Merge the " New classification models" pull request (#1657)

What I did:

  1. Locally, I cloned master, and then merged feat/new_classifiers

  2. updated utils/Docker/docker-compose.yml to only have py311 in it.

  3. $env:EPOCH=$(Get-Date -UFormat %s); $env:TFHUB_CACHE_DIR="tfhub_cache"; docker-compose -f utils/Docker/docker-compose.yml build

  4. docker compose -f utils/Docker/docker-compose.yml up

However, somehow the image I’ve just built seems to be a test-focused image, and not giving me an image similar to what’s in DockerHub.

It doesn’t seem to be exposing the riseapi interface.

How exactly is the deeppavlov/deeppavlov Docker image built?

Many thanks!

I searched the whole deeppavlov organization for some keyword (PYTHON_GET_PIP_URL) that is in the deeppavlov/deeppavlov docker image, but got no results: Code search results · GitHub

This Dockerfile doesn’t seem to be what builds the DockerHub images either: stand_kubernetes_cluster/utils/dp_base/cpu/Dockerfile at master · deeppavlov/stand_kubernetes_cluster · GitHub

I can’t match the instructions between what’s in the hub, and what’s in this Dockerfile.

Is the Dockerfile, that builds deeppavlov/deeppavlov, not open-source?

Thank you very much.