Merge branch 'master' into fix-exception-chaining

.github/workflows/ci_test-base.yml

@@ -41,14 +41,14 @@ jobs:
       uses: actions/cache@v2
       with:
         path: ${{ steps.pip-cache.outputs.dir }}
-        key: ${{ runner.os }}-${{ matrix.python-version }}-${{ matrix.requires }}-pip-${{ hashFiles('requirements/base.txt') }}
+        key: ${{ runner.os }}-${{ matrix.python-version }}-${{ matrix.requires }}-pip-${{ hashFiles('requirements.txt') }}
         restore-keys: |
           ${{ runner.os }}-${{ matrix.python-version }}-${{ matrix.requires }}-pip-
 
     - name: Install dependencies
       run: |
         python -m pip install --upgrade --user pip
-        pip install --requirement ./requirements/base.txt --quiet  --find-links https://download.pytorch.org/whl/cpu/torch_stable.html --upgrade
+        pip install --requirement ./requirements.txt --quiet  --find-links https://download.pytorch.org/whl/cpu/torch_stable.html --upgrade
         pip install --requirement ./requirements/test.txt --quiet --upgrade-strategy only-if-needed
         # pip install tox coverage
         python --version
@@ -66,7 +66,7 @@ jobs:
     - name: Test Package [only]
       run: |
         # NOTE: run coverage on tests does not propagare faler status for Win, https://github.com/nedbat/coveragepy/issues/1003
-        coverage run --source pl_bolts -m pytest pl_bolts -v --junitxml=junit/test-results-${{ runner.os }}-${{ matrix.python-version }}-${{ matrix.requires }}.xml --ignore=pl_bolts/datamodules --ignore=pl_bolts/models/self_supervised/amdim/transforms.py
+        coverage run --source pl_bolts -m pytest pl_bolts -v --junitxml=junit/test-results-${{ runner.os }}-${{ matrix.python-version }}-${{ matrix.requires }}.xml --ignore=pl_bolts/datamodules --ignore=pl_bolts/models/self_supervised/amdim/transforms.py --ignore=pl_bolts/models/rl
 
     - name: Upload pytest test results
       uses: actions/upload-artifact@master

.github/workflows/ci_test-full.yml

@@ -45,7 +45,7 @@ jobs:
     - name: Set min. dependencies
       if: matrix.requires == 'minimal'
       run: |
-        python -c "fpath = 'requirements/base.txt' ; req = open(fpath).read().replace('>=', '==') ; open(fpath, 'w').write(req)"
+        python -c "fpath = 'requirements.txt' ; req = open(fpath).read().replace('>=', '==') ; open(fpath, 'w').write(req)"
         python -c "fpath = 'requirements/models.txt' ; req = open(fpath).read().replace('>=', '==') ; open(fpath, 'w').write(req)"
         python -c "fpath = 'requirements/loggers.txt' ; req = open(fpath).read().replace('>=', '==') ; open(fpath, 'w').write(req)"
         python -c "fpath = 'requirements/test.txt' ; req = open(fpath).read().replace('>=', '==') ; open(fpath, 'w').write(req)"
@@ -61,7 +61,7 @@ jobs:
       uses: actions/cache@v2
       with:
         path: ${{ steps.pip-cache.outputs.dir }}
-        key: ${{ runner.os }}-pip-py${{ matrix.python-version }}-${{ matrix.requires }}-${{ hashFiles('requirements/base.txt') }}-${{ hashFiles('requirements/modules.txt') }}
+        key: ${{ runner.os }}-pip-py${{ matrix.python-version }}-${{ matrix.requires }}-${{ hashFiles('requirements.txt') }}-${{ hashFiles('requirements/modules.txt') }}
         restore-keys: |
           ${{ runner.os }}-pip-py${{ matrix.python-version }}-${{ matrix.requires }}-
 

.github/workflows/code-format.yml

@@ -23,14 +23,14 @@ jobs:
         uses: actions/cache@v2
         with:
           path: ~/.cache/pip
-          key: ${{ runner.os }}-pip-${{ hashFiles('requirements/base.txt') }}
+          key: ${{ runner.os }}-pip-${{ hashFiles('requirements.txt') }}
           restore-keys: |
             ${{ runner.os }}-pip-
 
       - name: Install dependencies
         run: |
           # python -m pip install --upgrade --user pip
-          pip install -r requirements/base.txt -U -f https://download.pytorch.org/whl/torch_stable.html -q
+          pip install -r requirements.txt -U -f https://download.pytorch.org/whl/torch_stable.html -q
           pip install flake8
           python --version
           pip --version

.github/workflows/docs-check.yml

@@ -36,7 +36,7 @@ jobs:
 #        uses: actions/cache@v2
 #        with:
 #          path: ~/.cache/pip
-#          key: ${{ runner.os }}-pip-${{ hashFiles('requirements/base.txt') }}
+#          key: ${{ runner.os }}-pip-${{ hashFiles('requirements.txt') }}
 #          restore-keys: |
 #            ${{ runner.os }}-pip-
 #
@@ -75,13 +75,13 @@ jobs:
         uses: actions/cache@v2
         with:
           path: ~/.cache/pip
-          key: ${{ runner.os }}-pip-${{ hashFiles('requirements/base.txt') }}
+          key: ${{ runner.os }}-pip-${{ hashFiles('requirements.txt') }}
           restore-keys: |
             ${{ runner.os }}-pip-
 
       - name: Install dependencies
         run: |
-          pip install --requirement requirements/base.txt --upgrade-strategy only-if-needed --find-links https://download.pytorch.org/whl/cpu/torch_stable.html --quiet
+          pip install --requirement requirements.txt --upgrade-strategy only-if-needed --find-links https://download.pytorch.org/whl/cpu/torch_stable.html --quiet
           pip install --requirement docs/requirements.txt
           # install Texlive, see https://linuxconfig.org/how-to-install-latex-on-ubuntu-20-04-focal-fossa-linux
           sudo apt-get update && sudo apt-get install -y texlive-latex-extra dvipng texlive-pictures

.gitignore

@@ -138,7 +138,6 @@ MNIST
 
 # Lightning logs
 lightning_logs
-datasets
 *.gz
 *-batches-py
 simclr.py

CHANGELOG.md

@@ -31,6 +31,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Added Linear Regression
 - Added Moco2g
 - Added simclr
+- Added RL module
 - Added Loggers
 - Added Transforms
 - Added Tiny Datasets
@@ -42,12 +43,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
+- Device is no longer set in the DQN model init
+- Moved RL loss function to the losses module
+- Moved rl.common.experience to datamodules
 - train_batch function to VPG model to generate batch of data at each step (POC)
 - Experience source no longer gets initialized with a device, instead the device is passed at each step()
 - Refactored ExperienceSource classes to be handle multiple environments. 
 
 ### Removed
 
+- Removed N-Step DQN as the latest version of the DQN supports N-Step by setting the `n_step` arg to n
 - Deprecated common.experience
 
 ### Fixed

MANIFEST.in

@@ -25,6 +25,7 @@ recursive-exclude docs *
 exclude docs
 
 # Include the Requirements
+include requirements.txt
 recursive-include requirements *.txt
 
 # Exclude build configs

README.md

@@ -57,7 +57,7 @@ Install bleeding-edge (no guarantees)
 pip install git+https://github.com/PytorchLightning/pytorch-lightning-bolts.git@master --upgrade
 ```
 
-In case you wan to have full experience you can install all optional packages at once
+In case you want to have full experience you can install all optional packages at once
 ```bash
 pip install pytorch-lightning-bolts["extra"]
 ```

docs/source/classic_ml.rst

@@ -9,7 +9,7 @@ half-precision training.
 Linear Regression
 -----------------
 Linear regression fits a linear model between a real-valued target variable :math:`y` and one or more features :math:`X`. We
-estimate the regression coefficients that minimizes the mean squared error between the predicted and true target
+estimate the regression coefficients that minimize the mean squared error between the predicted and true target
 values.
 
 We formulate the linear regression model as a single-layer neural network. By default we include only one neuron in
@@ -69,7 +69,7 @@ Add either L1 or L2 regularization, or both, by specifying the regularization st
 
     trainer.test(test_dataloaders=dm.test_dataloader(batch_size=12))
 
-Any input will be flattened across all dimensions except the firs one (batch).
+Any input will be flattened across all dimensions except the first one (batch).
 This means images, sound, etc... work out of the box.
 
 .. code-block:: python

docs/source/conf.py

@@ -328,7 +328,7 @@ def package_list_from_file(file):
 MOCK_PACKAGES = []
 if SPHINX_MOCK_REQUIREMENTS:
     # mock also base packages when we are on RTD since we don't install them there
-    MOCK_PACKAGES += package_list_from_file(os.path.join(PATH_ROOT, 'requirements', 'base.txt'))
+    MOCK_PACKAGES += package_list_from_file(os.path.join(PATH_ROOT, 'requirements.txt'))
     MOCK_PACKAGES += package_list_from_file(os.path.join(PATH_ROOT, 'requirements', 'models.txt'))
     MOCK_PACKAGES += package_list_from_file(os.path.join(PATH_ROOT, 'requirements', 'loggers.txt'))
 

docs/source/dataloaders.rst

@@ -3,19 +3,14 @@ AsynchronousLoader
 This dataloader behaves identically to the standard pytorch dataloader, but will transfer
 data asynchronously to the GPU with training. You can also use it to wrap an existing dataloader.
 
-Example::
+Example:
+
+.. code-block:: python
+
     dataloader = AsynchronousLoader(DataLoader(ds, batch_size=16), device=device)
 
     for b in dataloader:
         ...
 
 .. autoclass:: pl_bolts.datamodules.async_dataloader.AsynchronousLoader
    :noindex:
-
-------------------
-
-DummyDataset
-------------
-
-.. autoclass:: pl_bolts.datamodules.dummy_dataset.DummyDataset
-   :noindex:

docs/source/datamodules.rst

@@ -7,9 +7,9 @@ DataModules (introduced in PyTorch Lightning 0.9.0) decouple the data from a mod
 is simply a collection of a training dataloder, val dataloader and test dataloader. In addition,
 it specifies how to:
 
-- Downloading/preparing data.
+- Download/prepare data.
 - Train/val/test splits.
-- Transforms
+- Transform
 
 Then you can use it like this:
 

docs/source/datasets.rst

@@ -0,0 +1,41 @@
+########
+Datasets
+########
+Collection of useful datasets
+
+--------
+
+*********
+Debugging
+*********
+Use these datasets to debug
+
+DummyDataset
+============
+
+.. autoclass:: pl_bolts.datasets.dummy_dataset.DummyDataset
+    :noindex:
+
+DummyDetectionDataset
+=====================
+
+.. autoclass:: pl_bolts.datasets.dummy_dataset.DummyDetectionDataset
+    :noindex:
+
+RandomDataset
+=============
+
+.. autoclass:: pl_bolts.datasets.dummy_dataset.RandomDataset
+    :noindex:
+
+RandomDictDataset
+=================
+
+.. autoclass:: pl_bolts.datasets.dummy_dataset.RandomDictDataset
+    :noindex:
+
+RandomDictStringDataset
+=======================
+
+.. autoclass:: pl_bolts.datasets.dummy_dataset.RandomDictStringDataset
+    :noindex:

docs/source/index.rst

@@ -33,6 +33,13 @@ PyTorch-Lightning-Bolts documentation
    sklearn_datamodule
    vision_datamodules
 
+.. toctree::
+   :maxdepth: 2
+   :name: datasets
+   :caption: Datasets
+
+   datasets
+
 .. toctree::
    :maxdepth: 2
    :name: dataloaders
@@ -53,10 +60,17 @@ PyTorch-Lightning-Bolts documentation
    :caption: Models
 
    models_howto
-   autoencoders
    classic_ml
+
+.. toctree::
+   :maxdepth: 2
+   :name: vision
+   :caption: Vision models
+
+   autoencoders
    convolutional
    gans
+   reinforce_learn
    self_supervised_models
 
 .. toctree::
@@ -90,6 +104,7 @@ Indices and tables
    readme
    api/pl_bolts.callbacks
    api/pl_bolts.datamodules
+   api/pl_bolts.datasets
    api/pl_bolts.metrics
    api/pl_bolts.models
    api/pl_bolts.callbacks

docs/source/introduction_guide.rst

@@ -10,7 +10,7 @@ Bolts is a Deep learning research and production toolbox of:
 - Losses.
 - Datasets.
 
-**The Main goal of bolts is to enable trying new ideas as fast as possible!**
+**The Main goal of Bolts is to enable trying new ideas as fast as possible!**
 
 All models are tested (daily), benchmarked, documented and work on CPUs, TPUs, GPUs and 16-bit precision.
 
@@ -90,11 +90,11 @@ All models are tested (daily), benchmarked, documented and work on CPUs, TPUs, G
 
 Community Built
 ---------------
-Bolts are built-by the Lightning community and contributed to bolts.
+Then lightning community builds bolts and contributes them to Bolts.
 The lightning team guarantees that contributions are:
 
-1. Rigorously Tested (CPUs, GPUs, TPUs).
-2. Rigorously Documented.
+1. Rigorously tested (CPUs, GPUs, TPUs).
+2. Rigorously documented.
 3. Standardized via PyTorch Lightning.
 4. Optimized for speed.
 5. Checked for correctness.
@@ -351,7 +351,7 @@ In case your job or research doesn't need a "hammer", we offer implementations o
 which benefit from lightning's multi-GPU and TPU support.
 
 So, now you can run huge workloads scalably, without needing to do any engineering.
-For instance, here we can run Logistic Regression on Imagenet (each epoch takes about 3 minutes)!
+For instance, here we can run logistic Regression on Imagenet (each epoch takes about 3 minutes)!
 
 .. code-block:: python
 
@@ -414,7 +414,7 @@ But more importantly, you can scale up to many GPUs, TPUs or even CPUs
 
 Logistic Regression
 ^^^^^^^^^^^^^^^^^^^
-Here's an example for Logistic regression
+Here's an example for logistic regression
 
 .. code-block:: python
 
@@ -436,7 +436,7 @@ Here's an example for Logistic regression
 
     trainer.test(test_dataloaders=dm.test_dataloader(batch_size=12))
 
-Any input will be flattened across all dimensions except the firs one (batch).
+Any input will be flattened across all dimensions except the first one (batch).
 This means images, sound, etc... work out of the box.
 
 .. code-block:: python

docs/source/losses.rst

@@ -10,3 +10,33 @@ This package lists common losses across research domains
 Your Loss
 ---------
 We're cleaning up many of our losses, but in the meantime, submit a PR to add your loss here!
+
+-------------
+
+Reinforcement Learning
+======================
+These are common losses used in RL.
+
+---------------
+
+DQN Loss
+--------
+
+.. autofunction:: pl_bolts.losses.rl.dqn_loss
+    :noindex:
+
+---------------
+
+Double DQN Loss
+---------------
+
+.. autofunction:: pl_bolts.losses.rl.double_dqn_loss
+    :noindex:
+
+---------------
+
+Per DQN Loss
+------------
+
+.. autofunction:: pl_bolts.losses.rl.per_dqn_loss
+    :noindex:

docs/source/models.rst

@@ -15,7 +15,7 @@ by adding your contribution to bolts you get these **additional** benefits!
     6. We'll pretrain expensive models for you and host weights.
     7. We will improve the speed of your models!
     8. Eligible for invited talks to discuss your implementation.
-    9. Lightning Swag + involvement in the broader contributor community :)
+    9. Lightning swag + involvement in the broader contributor community :)
 
 .. note:: You still get to keep your attribution and be recognized for your work!
 
@@ -98,7 +98,7 @@ We request that each contribution have:
     - Your name and your team's name as the implementation authors.
     - Your team's affiliation
     - Any generated examples, or result plots.
-    - Hyperparameters configurations for the results.
+    - Hyperparameter configurations for the results.
 
 Thank you for all your amazing contributions!