Update documentation for PytorchExperiment and Config

LICENSE

@@ -1,6 +1,6 @@
 The MIT License (MIT)
 
-Copyright (c) 2017 Medical Image Computing Group, DKFZ
+Copyright (c) 2018 Medical Image Computing Group, DKFZ
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal
@@ -19,4 +19,3 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.
-

Makefile

@@ -1,8 +1,3 @@
-init:
-	pip install numpy ; \
-  pip install -f http://www.simpleitk.org/SimpleITK/resources/software.html --trusted-host www.simpleitk.org -r requirements.txt
-
-
 tests:
 	python -m unittest discover
 

Readme.md

@@ -8,7 +8,7 @@
 </p>
 
 Finally get some structure into your machine learning experiments.
-**trixi** is a tool that helps you configure, log and visualize your experiments in a reproducible fashion.
+**trixi** (Training & Retrospective Insights eXperiment Infrastructure) is a tool that helps you configure, log and visualize your experiments in a reproducible fashion.
 
 * [Features](#features)
 * [Installation](#installation)
@@ -28,7 +28,7 @@ Finally get some structure into your machine learning experiments.
 * Experiment Browser <br>
     *Compare, combine and visually inspect the results of your experiments*.
 
-An detailed implementation overview is given [here](https://trixi.readthedocs.io/en/latest/class_diagram.html).
+An implementation diagram is given [here](https://trixi.readthedocs.io/en/latest/class_diagram.html).
 
 ### Logging API
 
@@ -64,7 +64,7 @@ Here are some examples:
 ### Experiment Infrastructure
 
 The [Experiment Infrastructure](https://trixi.readthedocs.io/en/latest/_api/trixi.experiment.html) provides a unified way to configure, run, store and evaluate your results.
-It gives you an Experiment interface, for which you can implement the training, validation and testing.
+It gives you an experiment interface, for which you can implement the training, validation and testing.
 Furthermore it automatically provides you with easy access to the Logging API and stores your config as well as the
 results for easy evaluation and reproduction. There is an abstract [Experiment](https://trixi.readthedocs.io/en/latest/_api/trixi.experiment.html#trixi.experiment.experiment.Experiment) class and a [PytorchExperiment](https://trixi.readthedocs.io/en/latest/_api/trixi.experiment.html#trixi.experiment.pytorchexperiment.PytorchExperiment) with many convenience features.
 
@@ -73,6 +73,9 @@ results for easy evaluation and reproduction. There is an abstract [Experiment](
 For more info, visit the [Documentation](https://trixi.readthedocs.io/en/latest/_api/trixi.experiment.html).
 
 ### Experiment Browser
+
+**(We're currently remaking this from scratch, expect major improvements :))**
+
 The Experiment Browser offers a complete overview of experiments along with all config parameters and results.
 It also allows to combine and/or compare different experiments, giving you an interactive comparison highlighting differences in the configs and a detailed view of all images,
 plots, results and logs of each experiment, with live plots and more.

doc/index.rst

@@ -22,7 +22,7 @@ You can jump right into the package by looking into our :ref:`quick-start-label`
    quickstart
    license
    authors
-   changes
+   Github <https://github.com/MIC-DKFZ/trixi>
 
 .. toctree::
    :maxdepth: 1

requirements_full.txt

@@ -1,4 +1,4 @@
 -r requirements.txt
 python-telegram-bot==10.1.0
-umap>=0.1.1
+umap-learn>=0.3.6
 scikit-learn>=0.19.1

trixi/experiment/experiment.py

@@ -16,27 +16,31 @@ class Experiment(object):
 
         end()
 
+    The reason there is both :meth:`.setup` and :meth:`.prepare` is that internally there is also
+    a :meth:`._setup_internal` method for hidden magic in classes that inherit from this. For
+    example, the :class:`trixi.experiment.pytorchexperiment.PytorchExperiment` uses this to restore checkpoints. Think
+    if :meth:`.setup` as an :meth:`.__init__` that is only called when the Experiment is actually
+    asked to do anything. Then use :meth:`.prepare` to modify the fully instantiated Experiment if
+    you need to.
+
     To write a new Experiment simply inherit the Experiment class and overwrite the methods.
-    You can then start your experiment calling :meth:`.run`
+    You can then start your Experiment calling :meth:`.run`
 
-    In Addition the Experiment also has a test function. If you call the :meth:`.run_test` method is will call the
-    :meth:`.test` and :meth:`.end_test` method internally (and if you give the parameter setup = True
-    in run_test is will again call :meth:`.setup` and :meth:`.prepare` ).
+    In Addition the Experiment also has a test function. If you call the :meth:`.run_test` method it
+    will call the :meth:`.test` and :meth:`.end_test` method internally (and if you give the
+    parameter setup = True in run_test is will again call :meth:`.setup` and :meth:`.prepare` ).
 
-    Each experiment also has its current state in  :attr:`_exp_state`, its start time in  :attr:`_time_start`,
-    its end time in :attr:`_time_end` and the current epoch index in :attr:`_epoch_idx`
+    Each Experiment also has its current state in  :attr:`_exp_state`, its start time in
+    :attr:`_time_start`, its end time in :attr:`_time_end` and the current epoch index in
+    :attr:`_epoch_idx`
 
-    """
+    Args:
+        n_epochs (int): The number of epochs in the Experiment (how often the train and validate
+            method will be called)
 
+    """
 
     def __init__(self, n_epochs=0):
-        """
-        Initializes a new Experiment with a given number of epochs
-
-        Args:
-            n_epochs (int): The number of epochs in the experiment (how often the train and validate method
-                will be called)
-        """
 
         self.n_epochs = n_epochs
         self._exp_state = "Preparing"
@@ -46,7 +50,7 @@ def __init__(self, n_epochs=0):
 
     def run(self):
         """
-        This method runs the experiment. It runs through the basic lifecycle of an experiment::
+        This method runs the Experiment. It runs through the basic lifecycle of an Experiment::
 
             setup()
             prepare()
@@ -99,7 +103,7 @@ def run(self):
 
     def run_test(self, setup=True):
         """
-        This method runs the experiment.
+        This method runs the Experiment.
 
         The test consist of an optional setup and then calls the :meth:`.test` and :meth:`.end_test`.
 
@@ -137,12 +141,12 @@ def run_test(self, setup=True):
             raise e
 
     def setup(self):
-        """Is called at the beginning of each experiment run to setup the basic components needed for a run"""
+        """Is called at the beginning of each Experiment run to setup the basic components needed for a run"""
         pass
 
     def train(self, epoch):
         """
-        The training part of the experiment, it is called once for each epoch
+        The training part of the Experiment, it is called once for each epoch
 
         Args:
             epoch (int): The current epoch the train method is called in
@@ -152,7 +156,7 @@ def train(self, epoch):
 
     def validate(self, epoch):
         """
-        The evaluation/validation part of the experiment, it is called once for each epoch (after the training
+        The evaluation/validation part of the Experiment, it is called once for each epoch (after the training
         part)
 
         Args:
@@ -162,7 +166,7 @@ def validate(self, epoch):
         pass
 
     def test(self):
-        """The testing part of the experiment"""
+        """The testing part of the Experiment"""
         pass
 
     def process_err(self, e):