OS-specific installation (#235)

* doc: split macos and linux cpu installation

* doc: update

* ci: ident removal

* Update CONTRIBUTING.md

Co-authored-by: Matteo Manica <[email protected]>

* Update README.md

Co-authored-by: Matteo Manica <[email protected]>

* Update docs/source/gt4sd_inference_usage_md.md

Co-authored-by: Matteo Manica <[email protected]>

---------

Co-authored-by: Matteo Manica <[email protected]>

.github/workflows/docs.yaml

@@ -19,7 +19,7 @@ jobs:
       - uses: conda-incubator/setup-miniconda@v2
         with:
           activate-environment: gt4sd
-          environment-file: conda.yml
+          environment-file: conda_cpu_linux.yml
           auto-activate-base: false
           use-only-tar-bz2: true
       - name: Build docs

.github/workflows/tests.yaml

@@ -24,7 +24,7 @@ jobs:
       - uses: conda-incubator/setup-miniconda@v2
         with:
           activate-environment: gt4sd
-          environment-file: conda.yml
+          environment-file: conda_cpu_linux.yml
           auto-activate-base: false
           use-only-tar-bz2: true
       - name: Install gt4sd from source

CONTRIBUTING.md

@@ -48,7 +48,7 @@ If you would like to contribute to the package, we recommend the following devel
 5. Create and activate a dedicated conda environment:
 
     ```sh
-    conda env create -f conda.yml
+    conda env create -f conda_cpu_mac.yml # for linux use conda_cpu_linux.yml
     conda activate gt4sd
     ```
 

README.md

@@ -39,12 +39,13 @@ The recommended way to install the `gt4sd` is to create a dedicated conda enviro
 ```sh
 git clone https://github.com/GT4SD/gt4sd-core.git
 cd gt4sd-core/
-conda env create -f conda.yml 
+conda env create -f conda_cpu_mac.yml # for linux use conda_cpu_linux.yml
 conda activate gt4sd
 pip install gt4sd
 ```
 
-**NOTE 1:** By default `gt4sd` is installed with CPU requirements. For GPU usage replace `conda env create -f conda.yml` with:
+**NOTE 1:** By default `gt4sd` is installed with CPU requirements. For GPU usage replace with:
+
 ```sh
 conda env create -f conda_gpu.yml
 ```
@@ -59,12 +60,10 @@ A few VCS dependencies require Git LFS (make sure it's available on your system)
 
 ### Development setup & installation
 
-If you would like to contribute to the package, we recommend the following development setup:
+If you would like to contribute to the package, we recommend to install gt4sd in
+editable mode inside your `conda` environment:
 
 ```sh
-conda env create -f conda.yml
-conda activate gt4sd
-# install gt4sd in editable mode
 pip install --no-deps -e .
 ```
 

conda_cpu_mac.yml

@@ -0,0 +1,18 @@
+name: gt4sd
+channels:
+  - https://conda.anaconda.org/pytorch
+  - https://conda.anaconda.org/nvidia
+  - https://conda.anaconda.org/pyg
+  - https://conda.anaconda.org/conda-forge
+dependencies:
+  - python>=3.7,<3.9
+  - pip>=19.1,<20.3
+  - pytorch>=1.0,<=1.12.1
+  - cpuonly
+  - pytorch-scatter<=2.0.9
+  - pip:
+      - -r requirements.txt
+      - -r cpu_requirements.txt
+      - -r vcs_requirements.txt
+      # development
+      - -r dev_requirements.txt

docs/source/gt4sd_inference_usage_md.md

@@ -17,11 +17,11 @@ This notebook show the basic usage of the GT4SD algorithms.
 
 ### A note on the setup
 
-In the following we assume that the toolkit has been setup using the provided `conda.yml` as follows:
+In the following we assume that the toolkit has been setup using your preferred environment file (`conda*.yml`):
 
 ```{code-block} sh
 # create and activate environment
-conda env create -f conda.yml
+conda env create -f conda_cpu_mac.yml # or conda_cpu_linux.yml or conda_gpu.yml
 conda activate gt4sd
 # install the toolkit
 pip install .