Release: v1.2.6 and updated installation requirements/instructions.

README.rst

@@ -24,7 +24,7 @@ Here, we describe how to install osl-dynamics from source. We recommend using th
     conda activate osld
     pip install -e .
 
-If you're using a Mac then use the ``envs/mac.yml`` file to create the conda environment instead of ``envs/linux.yml``. Note, the conda environments use ``pip`` to install TensorFlow, you may need to load/install additional libraries (such as CUDA/cuDNN) if you have GPU support.
+Note, if you have a Mac you may want to use the ``envs/mac.yml`` environment file instead.
 
 Developers might want to clone the repo using SSH instead of HTTPS:
 
@@ -46,7 +46,8 @@ If you have already installed `OSL <https://github.com/OHBA-analysis/osl>`_ you
 
     conda activate osl
     cd osl-dynamics
-    pip install tensorflow
+    pip install tensorflow==2.9.1
+    pip install tensorflow-probability==0.17
     pip install -e .
 
 Documentation

doc/documentation.rst

@@ -5,7 +5,7 @@ Welcome to the osl-dynamics documentation!
 
 The :doc:`API reference <autoapi/index>` provides documentation for the classes, methods and functions in osl-dynamics. New users may find the :doc:`FAQ <faq>` useful.
 
-**New users are encouraged to have a look at the config API inferface:** `here <https://osl-dynamics.readthedocs.io/en/latest/autoapi/osl_dynamics/config_api/index.html>`_.
+**New users are encouraged to have a look at the config API interface:** `here <https://osl-dynamics.readthedocs.io/en/latest/autoapi/osl_dynamics/config_api/index.html>`_.
 
 Models
 ------

doc/install.rst

@@ -17,25 +17,29 @@ OSL Dynamics can be installed in three steps. Open a Terminal and execute the fo
 
     Note, this environment must be activated every time you want to use osl-dynamics.
 
-#. Install the deep learning library TensorFlow: https://www.tensorflow.org/overview.
+#. Install the deep learning library TensorFlow: https://www.tensorflow.org/overview (and the addon tensorflow-probability).
+
+    To install TensorFlow use:
 
     ::
 
-        pip install tensorflow
+        pip install tensorflow==2.9.1
+
+    If you have GPU resources you may need to install additional libraries (CUDA/cuDNN), see https://www.tensorflow.org/install/pip for detailed instructions.
 
-    If you are using an Apple computer with an M1/M2 chip the above command won't work, instead you can install TensorFlow with:
+    If you are using an Apple computer with an M1/M2 chip, you will need to use the following instead:
 
     ::
 
-        pip install tensorflow-macos
+        pip install tensorflow-macos==2.9.1
 
-    If you have GPU resources you need to install additional libraries (CUDA/cuDNN), see https://www.tensorflow.org/install/pip for detailed instructions. You maybe able you install a GPU-enabled version of TensorFlow using Anaconda:
+    After you have installed TensorFlow, install the tensorflow-probability addon with:
 
     ::
 
-        conda install -c conda-forge tensorflow-gpu
+        pip install tensorflow-probability==0.17
 
-#. Install osl-dynamics:
+#. Finally, install osl-dynamics:
 
     ::
 

envs/linux.yml

@@ -1,6 +1,6 @@
 name: osld
 dependencies:
-  - python=3
+  - python=3.10
   - numba
   - pip
   - pip:
@@ -18,6 +18,6 @@ dependencies:
     - scikit-learn
     - seaborn
     - tabulate
-    - tensorflow
-    - tensorflow_probability
     - tqdm
+    - tensorflow==2.9.1
+    - tensorflow-probability==0.17

envs/mac.yml

@@ -1,6 +1,6 @@
 name: osld
 dependencies:
-  - python=3
+  - python=3.10
   - numba
   - pip
   - pip:
@@ -18,6 +18,6 @@ dependencies:
     - scikit-learn
     - seaborn
     - tabulate
-    - tensorflow-macos
-    - tensorflow_probability
     - tqdm
+    - tensorflow-macos==2.9.1
+    - tensorflow-probability==0.17

setup.cfg

@@ -2,7 +2,7 @@
 
 [metadata]
 name = osl-dynamics
-version = 1.2.dev6
+version = 1.2.6
 description = Models for infering dynamics in neuroimaging data
 license = MIT
 long_description = file: README.rst
@@ -28,7 +28,6 @@ install_requires =
     scikit-learn
     seaborn
     tabulate
-    tensorflow_probability
     tqdm
 packages =
     osl_dynamics