Add configs

projects/JSSL/README.rst

@@ -1,6 +1,10 @@
 JSSL: Joint Supervised and Self-supervised Learning for MRI Reconstruction
 ==========================================================================
 
+This project contains necessary configuration files to reproduce experiments of the paper
+`JSSL: Joint Supervised and Self-supervised Learning for MRI Reconstruction <https://arxiv.org/abs/2311.15856>`_.
+
+
 .. figure:: https://github.com/NKI-AI/direct/assets/71031687/97f1fa27-f4e4-44e2-b54a-7ac149f3c01d
    :alt: fig
    :name: fig1
@@ -11,10 +15,6 @@ JSSL: Joint Supervised and Self-supervised Learning for MRI Reconstruction
    (2) Self-supervised Learning.
 
 
-This project contains necessary configuration files to reproduce experiments of the paper
-`JSSL: Joint Supervised and Self-supervised Learning for MRI Reconstruction <https://arxiv.org/abs/2311.15856>`_.
-
-
 Setting up data directory
 -------------------------
 
@@ -73,4 +73,38 @@ Filenames lists
 Provided configuration files assume that `direct/projects/JSSL/lists/` contains `.lst` files for each dataset
 which contains the list of filenames for training, validation and test datasets. Each list categorizes the filenames
 based on the dataset they belong to, as well as shape and number of coils. This is necessary for the dataloader to
-collate the data for batch size greater than 1. The filenames should be the same as the ones in the data directory. 
+collate the data for batch size greater than 1. The filenames should be the same as the ones in the data directory. 
+
+Experiments
+-----------
+
+Configuration Files
+~~~~~~~~~~~~~~~~~~~
+
+We provide configuration files for JSSL, SSL and all ablation settings as presented in the main paper, 
+as well as the supplementary material. The configuration files are stored
+in the `JSSL projects folder <https://github.com/NKI-AI/direct/tree/main/projects/JSSL>`_.
+
+
+Training
+~~~~~~~~
+
+In `direct/` run the following command to commence training:
+
+.. code-block:: bash
+
+    direct train <output_folder> \
+      --training-root <base_path>/training/ \
+      --validation-root <base_path>/validation/ \
+      --cfg projects/JSSL/configs/<name_of_experiment>.yaml \
+      --num-gpus <number_of_gpus> \
+      --num-workers <number_of_workers> --resume
+
+
+For instance, to perform JSSL training with the vSHARP model, replace `<name_of_experiment>.yaml`
+with `vsharp_jssl.yaml`.
+
+Note
+~~~~
+Note that `jssl`, `sl` and `ssl` in the configuration file names stand for Joint Supervised and Self-supervised 
+Learning, Supervised Learning, and Self-supervised Learning, respectively.