Can't build docker image because catboost won't install

[VikramTiwari]

Logs:

$ docker build -t tabsurvey .
[+] Building 1.1s (18/49)                                                                                                                                                                    
 => [internal] load build definition from Dockerfile                                                                                                                                    0.0s
 => => transferring dockerfile: 37B                                                                                                                                                     0.0s
 => [internal] load .dockerignore                                                                                                                                                       0.0s
 => => transferring context: 2B                                                                                                                                                         0.0s
 => [internal] load metadata for docker.io/continuumio/miniconda3:latest                                                                                                                0.5s
 => [ 1/46] FROM docker.io/continuumio/miniconda3@sha256:977263e8d1e476972fddab1c75fe050dd3cd17626390e874448bd92721fd659b                                                               0.0s
 => CACHED [ 2/46] RUN /opt/conda/bin/conda install jupyter -y                                                                                                                          0.0s
 => CACHED [ 3/46] RUN mkdir /opt/notebooks                                                                                                                                             0.0s
 => CACHED [ 4/46] RUN opt/conda/bin/jupyter notebook --generate-config                                                                                                                 0.0s
 => CACHED [ 5/46] RUN /opt/conda/bin/conda create -n sklearn -y scikit-learn                                                                                                           0.0s
 => CACHED [ 6/46] RUN /opt/conda/bin/conda install -n sklearn -y -c anaconda ipykernel                                                                                                 0.0s
 => CACHED [ 7/46] RUN /opt/conda/envs/sklearn/bin/python -m ipykernel install --user --name=sklearn                                                                                    0.0s
 => CACHED [ 8/46] RUN /opt/conda/bin/conda install -n sklearn -y -c conda-forge optuna                                                                                                 0.0s
 => CACHED [ 9/46] RUN /opt/conda/bin/conda install -n sklearn -y -c conda-forge configargparse                                                                                         0.0s
 => CACHED [10/46] RUN /opt/conda/bin/conda install -n sklearn -y pandas                                                                                                                0.0s
 => CACHED [11/46] RUN /opt/conda/bin/conda create -n gbdt -y                                                                                                                           0.0s
 => CACHED [12/46] RUN /opt/conda/bin/conda install -n gbdt -y -c anaconda ipykernel                                                                                                    0.0s
 => CACHED [13/46] RUN /opt/conda/envs/gbdt/bin/python -m ipykernel install --user --name=gbdt                                                                                          0.0s
 => CACHED [14/46] RUN /opt/conda/envs/gbdt/bin/python -m pip install xgboost==1.5.0                                                                                                    0.0s
 => ERROR [15/46] RUN /opt/conda/envs/gbdt/bin/python -m pip install catboost==1.0.3                                                                                                    0.6s
------
 > [15/46] RUN /opt/conda/envs/gbdt/bin/python -m pip install catboost==1.0.3:
#18 0.547 ERROR: Could not find a version that satisfies the requirement catboost==1.0.3 (from versions: none)
#18 0.547 ERROR: No matching distribution found for catboost==1.0.3
------
executor failed running [/bin/sh -c /opt/conda/envs/gbdt/bin/python -m pip install catboost==1.0.3]: exit code: 1

Let me know if there's anything else I should add here.

[VikramTiwari]

I disabled catboost dependency and tried to build again and I am getting another error:

docker build -t tabsurvey .
[+] Building 2.9s (23/48)                                                                                                                                                                    
 => [internal] load build definition from Dockerfile                                                                                                                                    0.0s
 => => transferring dockerfile: 37B                                                                                                                                                     0.0s
 => [internal] load .dockerignore                                                                                                                                                       0.0s
 => => transferring context: 2B                                                                                                                                                         0.0s
 => [internal] load metadata for docker.io/continuumio/miniconda3:latest                                                                                                                0.4s
 => [ 1/45] FROM docker.io/continuumio/miniconda3@sha256:977263e8d1e476972fddab1c75fe050dd3cd17626390e874448bd92721fd659b                                                               0.0s
 => CACHED [ 2/45] RUN /opt/conda/bin/conda install jupyter -y                                                                                                                          0.0s
 => CACHED [ 3/45] RUN mkdir /opt/notebooks                                                                                                                                             0.0s
 => CACHED [ 4/45] RUN opt/conda/bin/jupyter notebook --generate-config                                                                                                                 0.0s
 => CACHED [ 5/45] RUN /opt/conda/bin/conda create -n sklearn -y scikit-learn                                                                                                           0.0s
 => CACHED [ 6/45] RUN /opt/conda/bin/conda install -n sklearn -y -c anaconda ipykernel                                                                                                 0.0s
 => CACHED [ 7/45] RUN /opt/conda/envs/sklearn/bin/python -m ipykernel install --user --name=sklearn                                                                                    0.0s
 => CACHED [ 8/45] RUN /opt/conda/bin/conda install -n sklearn -y -c conda-forge optuna                                                                                                 0.0s
 => CACHED [ 9/45] RUN /opt/conda/bin/conda install -n sklearn -y -c conda-forge configargparse                                                                                         0.0s
 => CACHED [10/45] RUN /opt/conda/bin/conda install -n sklearn -y pandas                                                                                                                0.0s
 => CACHED [11/45] RUN /opt/conda/bin/conda create -n gbdt -y                                                                                                                           0.0s
 => CACHED [12/45] RUN /opt/conda/bin/conda install -n gbdt -y -c anaconda ipykernel                                                                                                    0.0s
 => CACHED [13/45] RUN /opt/conda/envs/gbdt/bin/python -m ipykernel install --user --name=gbdt                                                                                          0.0s
 => CACHED [14/45] RUN /opt/conda/envs/gbdt/bin/python -m pip install xgboost==1.5.0                                                                                                    0.0s
 => CACHED [15/45] RUN /opt/conda/envs/gbdt/bin/python -m pip install lightgbm==3.3.1                                                                                                   0.0s
 => CACHED [16/45] RUN /opt/conda/bin/conda install -n gbdt -y -c conda-forge optuna                                                                                                    0.0s
 => CACHED [17/45] RUN /opt/conda/bin/conda install -n gbdt -y -c conda-forge configargparse                                                                                            0.0s
 => CACHED [18/45] RUN /opt/conda/bin/conda install -n gbdt -y pandas                                                                                                                   0.0s
 => CACHED [19/45] RUN /opt/conda/envs/gbdt/bin/python -m pip install https://github.com/schufa-innovationlab/model-trees/archive/master.zip                                            0.0s
 => ERROR [20/45] RUN /opt/conda/bin/conda create -n torch -y pytorch cudatoolkit=11.3 -c pytorch                                                                                       2.4s
------                                                                                                                                                                                       
 > [20/45] RUN /opt/conda/bin/conda create -n torch -y pytorch cudatoolkit=11.3 -c pytorch:                                                                                                  
#23 0.267 Collecting package metadata (current_repodata.json): ...working... done                                                                                                            
#23 0.538 Solving environment: ...working... failed with repodata from current_repodata.json, will retry with next repodata source.                                                          
#23 0.539 Collecting package metadata (repodata.json): ...working... done                                                                                                                    
#23 2.183 Solving environment: ...working... failed                                                                                                                                          
#23 2.184 
#23 2.184 PackagesNotFoundError: The following packages are not available from current channels:
#23 2.184 
#23 2.184   - cudatoolkit=11.3
#23 2.184 
#23 2.184 Current channels:
#23 2.184 
#23 2.184   - https://conda.anaconda.org/pytorch/linux-aarch64
#23 2.184   - https://conda.anaconda.org/pytorch/noarch
#23 2.184   - https://repo.anaconda.com/pkgs/main/linux-aarch64
#23 2.184   - https://repo.anaconda.com/pkgs/main/noarch
#23 2.184   - https://repo.anaconda.com/pkgs/r/linux-aarch64
#23 2.184   - https://repo.anaconda.com/pkgs/r/noarch
#23 2.184 
#23 2.184 To search for alternate channels that may provide the conda package you're
#23 2.184 looking for, navigate to
#23 2.184 
#23 2.184     https://anaconda.org
#23 2.184 
#23 2.184 and use the search bar at the top of the page.
#23 2.184 
#23 2.184 
------
executor failed running [/bin/sh -c /opt/conda/bin/conda create -n torch -y pytorch cudatoolkit=11.3 -c pytorch]: exit code: 1

[unnir]

Hey,

just tested docker build -t tabsurvey . CatBoost and PyTorch (with the cudatoolkit) are installed without any errors.

I suspect the issue is on your side.

[andyrosa2]

Same problem yesterday.
I changed Dockerfile to
RUN /opt/conda/envs/gbdt/bin/python -m pip install catboost==1.2
Also added this line after # For NAM
ENV SKLEARN_ALLOW_DEPRECATED_SKLEARN_PACKAGE_INSTALL=True
Their paper just got published on IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, VOL. 35, NO. 6, JUNE 2024.
https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9998482
This repo might be getting some attention soon.
I think the paper and the repo are very useful. It might use some TLC to get it updated.