Update docs for config system

Merge pull request #42 from aayushk614/patch-3

docs/source/notes/config.rst

@@ -1,4 +1,115 @@
 Configurations
 ===============
 
-We will add detailed guidance for using the configuration system soon!
+Connectomics uses a key-value based configuration system that can be used to carry outstandard and commonly used tasks.
+
+The config system of Connectomics uses YAML, a human-readable data-serialization language and yacs package, a simple experiment 
+configuration system for research. In addition to the general tasks, that access and update a config, we provide the following extra 
+functionalities:
+
+#. The config have ``_C.key:value``  field, which will use a pre-defined config first. Values in the pre-defined config will 
+   be overwritten in sub-configs, if there are any according to requirements. We provided several base configs for standard model 
+   architectures as ``task.yaml`` files.
+
+``config`` has a very limited abstraction. We do not expect all features in Connectomics to be available through configs. If you need 
+something that’s not available in the config space, please modify the keys-value pairs in the file at 
+``/connectomics/config/config.py``
+
+Basic Usage
+============
+
+Some basic usage of the ``CfgNode`` object is shown here.
+
+.. code-block:: none
+
+    from yacs.config import CfgNode as CN
+    _C = CN()            # config definition
+    _C.SYSTEM = CN()     # config definition for GPU and CPU
+    _C.MODEL = CN()      # Model architectures defined in the package
+    _C.MODEL.ARCHITECTURE = 'unet_residual_3d' 
+   
+The configs in Connectomics also accepts command line configuration overwrite, i.e.: Key-value pairs provided in the command line will 
+overwrite the existing values in the config file. For example, ``main.py`` can be used with to modify input file name :
+
+.. code-block:: none
+
+    python -u scripts/main.py \
+    --config-file configs/Lucchi-Mitochondria.yaml DATASET.IMAGE_NAME ‘img/train_im.h5’
+  
+To see a list of available configs in Connectomics and what they mean, check `Config References <https://github.com/zudi-
+lin/pytorch_connectomics/blob/master/connectomics/config/config.py>`_
+
+
+Best Practice with Configs
+==========================
+
+#. Treat the configs you write as “code”: avoid copying them or duplicating them; use ``_BASE_`` to share common parts between 
+configs.
+
+#. Keep the configs you write simple: don’t include keys that do not affect the experimental setting.
+
+
+Specifing multiple loss for a single learning target
+=========================================================
+
+We can also specify multiple loss functions as a criterion for training the segmentation models, i.e Binary cross-entropy loss, Dice 
+loss by changing the key-value pairs of ``_LOSS_OPTION`` in the ``config.yaml`` file or by giving it explicitly as command-line 
+arguments, multiple loss functions can be put to use for a single learning target.
+
+
+#. Say, you want to use Weighted Binary cross-entropy loss as well as Dice loss as criterions for mitochondria segmentation task.
+
+.. code-block:: none
+
+   python -u scripts/main.py --config-file configs/Lucchi-Mitochondria.yaml \
+   MODEL.LOSS_OPTION [[‘WeightedBCE’, ‘DiceLoss’]] MODEL.LOSS_WEIGHT [[1.0, 1.0]]
+   
+ 
+``LOSS_OPTION`` specifies the loss criterions to be used while training
+``LOSS_WEIGHT`` specifies the weight or emphasis to be given to each loss criterion, i.e In the above case both the loss criterion
+will contribute equally to the overall loss.
+
+
+#. Say, you want to use Weighted Binary cross-entropy loss as well as Dice loss as criterions with a weight of 1.0, 0.5 .
+
+.. code-block:: none
+
+   python -u scripts/main.py --config-file configs/Lucchi-Mitochondria.yaml \
+   MODEL.LOSS_OPTION [[‘WeightedBCE’, ‘DiceLoss’]] MODEL.LOSS_WEIGHT [[1.0, 0.5]]
+   
+ 
+``LOSS_OPTION`` specifies the loss criterions to be used while training
+``LOSS_WEIGHT`` specifies the weight or emphasis to be given to each loss criterion, i.e. In the above case cross-entropy will 
+contribute twice times more than dice loss to the overall loss.
+
+
+
+#. Say, you want to use only Weighted Binary cross-entropy loss as a criterion for mitochondria segmentation task.
+
+.. code-block:: none
+
+   python -u scripts/main.py --config-file configs/Lucchi-Mitochondria.yaml \
+   MODEL.LOSS_OPTION [[‘WeightedBCE’]] MODEL.LOSS_WEIGHT [[1.0]]
+   
+ 
+``LOSS_OPTION`` specifies only W Binary cross-entropy as loss criterions to be used while training.
+``LOSS_WEIGHT`` specifies the weight emphasis to be given to each loss criterion, i.e. In the above case, only cross-entropy will 
+contribute to the overall loss.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+