Merge branch 'master' into bugfix/3827_test_ddp_error

Author: SeanNaren

.drone.yml

@@ -25,13 +25,16 @@ steps:
   environment:
     CODECOV_TOKEN:
       from_secret: codecov_token
+    AUTH_TOKEN:
+      from_secret: gh_auth_token
     MKL_THREADING_LAYER: GNU
 
   commands:
     - python --version
     - pip --version
     - nvidia-smi
     - pip install -r ./requirements/devel.txt --upgrade-strategy only-if-needed -v --no-cache-dir
+    - pip install git+https://${AUTH_TOKEN}@github.com/PyTorchLightning/[email protected] -v --no-cache-dir
     # when Image has defined CUDa version we can switch to this package spec "nvidia-dali-cuda${CUDA_VERSION%%.*}0"
     # todo: temprarl fix till https://github.com/PyTorchLightning/pytorch-lightning/pull/4922 is resolved
     - pip install --extra-index-url https://developer.download.nvidia.com/compute/redist "nvidia-dali-cuda100<0.27" --upgrade-strategy only-if-needed

.github/CODEOWNERS

@@ -5,23 +5,48 @@
 # the repo. Unless a later match takes precedence,
 # @global-owner1 and @global-owner2 will be requested for
 # review when someone opens a pull request.
-* @williamfalcon @borda @teddykoker @awaelchli @nateraw @justusschock @tchaton @SeanNaren @ananyahjha93
+* @williamfalcon @borda @tchaton @SeanNaren @awaelchli @justusschock
 
 # Metrics
-/pytorch_lightning/metrics/*  @teddykoker @ananyahjha93 @justusschock
-/tests/metrics/*              @teddykoker @ananyahjha93 @justusschock
+/pytorch_lightning/metrics/   @teddykoker @ananyahjha93 @justusschock
+/tests/metrics/               @teddykoker @ananyahjha93 @justusschock
 /docs/source/metrics.rst      @teddykoker @ananyahjha93 @justusschock
 
 # API
-/pytorch_lightning/callbacks/base.py  @williamfalcon
-/pytorch_lightning/core/datamodule.py @williamfalcon
-/pytorch_lightning/trainer/trainer.py @williamfalcon
-/pytorch_lightning/core/hooks.py     @williamfalcon
-/pytorch_lightning/core/lightning.py  @williamfalcon
+/pytorch_lightning/callbacks/base.py           @williamfalcon
+/pytorch_lightning/core/datamodule.py          @williamfalcon
+/pytorch_lightning/trainer/trainer.py          @williamfalcon @tchaton
+/pytorch_lightning/core/hooks.py               @williamfalcon
+/pytorch_lightning/core/lightning.py           @williamfalcon @tchaton
+/pytorch_lightning/core/optimizer.py           @tchaton
+/pytorch_lightning/trainer/training_loop.py    @tchaton @SeanNaren
+/pytorch_lightning/trainer/evaluation_loop.py  @tchaton @SeanNaren
 
+# Connectors
+/pytorch_lightning/trainer/connectors/         @tchaton @SeanNaren
 
 # accelerators
-/pytorch_lightning/accelerators/*  @williamfalcon
+/pytorch_lightning/accelerators/               @williamfalcon @tchaton @SeanNaren @awaelchli @justusschock
 
 # owners
-/pytorch_lightning/.github/CODEOWNERS @williamfalcon
+/.github/CODEOWNERS   @williamfalcon
+# main
+/README.md            @williamfalcon @edenlightning
+# installation
+/setup.py             @borda @williamfalcon
+
+# CI/CD
+/.github/workflows/   @borda @tchaton
+/.github/*.py         @borda @tchaton
+/dockers/             @borda @tchaton
+# configs in root
+/*.yml                @borda @tchaton
+
+# Docs
+/docs/                          @edenlightning @tchaton @borda @awaelchli
+/.github/*.md                   @edenlightning @williamfalcon @borda
+/.github/ISSUE_TEMPLATE/*.md    @edenlightning @borda @tchaton
+/docs/source/conf.py            @borda @awaelchli
+
+# Testing
+/tests/base/boring_model.py @williamfalcon

.github/workflows/ci_test-full.yml

@@ -18,12 +18,9 @@ jobs:
         python-version: [3.6, 3.7, 3.8]
         requires: ['minimal', 'latest']
         exclude:
-          # excludes PT 1.3 as it is missing on pypi
+          # # todo: segmentation fault for minimal and hanging for latest
           - python-version: 3.8
-            requires: 'minimal'
-          # TODO: temporary fix till hanging jobs on macOS for py38 is resolved
-          - python-version: 3.8
-            os: macOS-10.15
+            os: ubuntu-18.04
 
     # Timeout: https://stackoverflow.com/a/59076067/4521646
     timeout-minutes: 35  # TODO: the macOS is taking too long, probably caching did not work...
@@ -36,7 +33,8 @@ jobs:
 
     - name: Update Pip
       run: |
-        pip install --quiet "pip>=20.1" --upgrade --user  # needed for get pip cacher folder
+        # todo: unfreeze PIP after resolving minimal dependencies
+        pip install --quiet "pip==20.1" --upgrade --user  # needed for get pip cacher folder
 
     # Github Actions: Run step on specific OS: https://stackoverflow.com/a/57948488/4521646
     - name: Setup macOS
@@ -52,16 +50,19 @@ jobs:
         python -c "fname = 'requirements/extra.txt' ; lines = [line for line in open(fname).readlines() if not line.startswith('horovod')] ; open(fname, 'w').writelines(lines)"
 
     # versions <= 1.3 may have issues on mac with some BLAS ops due to missing mkl (https://github.com/pytorch/pytorch/issues/18996)
-    - name: Setup MacOS Minimal
-      if: runner.os == 'macOS' && matrix.requires == 'minimal'
+    - name: Adjust minimal for Python 3.8 and MacOS
+      if: matrix.requires == 'minimal' && (runner.os == 'macOS' || matrix.python-version == 3.8)
       run : |
         python -c "fname = 'requirements.txt' ; req = open(fname).read().replace('torch>=1.3', 'torch>=1.4') ; open(fname, 'w').write(req)"
+        python -c "fname = 'requirements/examples.txt' ; req = open(fname).read().replace('torchvision>=0.4.1', 'torchvision>=0.5.0') ; open(fname, 'w').write(req)"
+        python -c "fname = 'requirements/extra.txt' ; req = open(fname).read().replace('torchtext>=0.3.1', 'torchtext>=0.5.0') ; open(fname, 'w').write(req)"
 
     - name: Set min. dependencies
       if: matrix.requires == 'minimal'
       run: |
         python -c "fname = 'requirements.txt' ; req = open(fname).read().replace('>=', '==') ; open(fname, 'w').write(req)"
         python -c "fname = 'requirements/extra.txt' ; req = open(fname).read().replace('>=', '==') ; open(fname, 'w').write(req)"
+        python -c "fname = 'requirements/loggers.txt' ; req = open(fname).read().replace('>=', '==') ; open(fname, 'w').write(req)"
         python -c "fname = 'requirements/test.txt' ; req = open(fname).read().replace('>=', '==') ; open(fname, 'w').write(req)"
         python -c "fname = 'requirements/examples.txt' ; req = open(fname).read().replace('>=', '==') ; open(fname, 'w').write(req)"
         # remove Fairscale from requirements

CHANGELOG.md

@@ -4,8 +4,14 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
+## Unreleased
+
+### Fixed
+
+- Fixed `LoggerConnector` to have logged metrics on root device in DP ([#4138](https://github.com/PyTorchLightning/pytorch-lightning/pull/4138))
 
-## [unreleased.Features] - YYYY-MM-DD
+
+## [1.1.0rc] - 2020-12-02
 
 ### Added
 
@@ -30,10 +36,12 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Added `current_score` to `ModelCheckpoint.on_save_checkpoint` ([#4721](https://github.com/PyTorchLightning/pytorch-lightning/pull/4721))
 
 
-- Added logging using `self.log` in train and evaluation for most callbacks and model hooks (
+- Added logging using `self.log` in train and evaluation for epoch end hooks (
     [#4552](https://github.com/PyTorchLightning/pytorch-lightning/pull/4552),
     [#4495](https://github.com/PyTorchLightning/pytorch-lightning/pull/4495),
     [#4439](https://github.com/PyTorchLightning/pytorch-lightning/pull/4439))
+    [#4684](https://github.com/PyTorchLightning/pytorch-lightning/pull/4684))
+    [#4913](https://github.com/PyTorchLightning/pytorch-lightning/pull/4913))
 
 
 - Added ability for DDP plugin to modify optimizer state saving ([#4675](https://github.com/PyTorchLightning/pytorch-lightning/pull/4675))
@@ -44,27 +52,75 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Added printing of total num of params, trainable and non-trainable params in ModelSummary ([#4521](https://github.com/PyTorchLightning/pytorch-lightning/pull/4521))
 
+
 - Added optimizer refactors ([#4658](https://github.com/PyTorchLightning/pytorch-lightning/pull/4658))
 
 
+- Added `PrecisionRecallCurve, ROC, AveragePrecision` class metric ([#4549](https://github.com/PyTorchLightning/pytorch-lightning/pull/4549))
+
+
+- Added custom `Apex` and `NativeAMP` as `Precision plugins` ([#4355](https://github.com/PyTorchLightning/pytorch-lightning/pull/4355))
+
+
+- Added `DALI MNIST` example ([#3721](https://github.com/PyTorchLightning/pytorch-lightning/pull/3721))
+
+
+- Added `sharded plugin` for DDP for multi-gpu training memory optimizations (
+                            [#4639](https://github.com/PyTorchLightning/pytorch-lightning/pull/4639),
+                            [#4686](https://github.com/PyTorchLightning/pytorch-lightning/pull/4686),
+                            [#4675](https://github.com/PyTorchLightning/pytorch-lightning/pull/4675),
+                            [#4737](https://github.com/PyTorchLightning/pytorch-lightning/pull/4737),
+                            [#4773](https://github.com/PyTorchLightning/pytorch-lightning/pull/4773))
+
+
+- Added `experiment_id` to the NeptuneLogger ([#3462](https://github.com/PyTorchLightning/pytorch-lightning/pull/3462))
+
+
+- Added `Pytorch Geometric` integration example with Lightning ([#4568](https://github.com/PyTorchLightning/pytorch-lightning/pull/4568))
+
+
 ### Changed
 
+- Removed `multiclass_roc` and `multiclass_precision_recall_curve`, use `roc` and `precision_recall_curve` instead ([#4549](https://github.com/PyTorchLightning/pytorch-lightning/pull/4549))
+
+
+
 - Tuner algorithms will be skipped if `fast_dev_run=True` ([#3903](https://github.com/PyTorchLightning/pytorch-lightning/pull/3903))
 
+
+
 - WandbLogger does not force wandb `reinit` arg to True anymore and creates a run only when needed ([#4648](https://github.com/PyTorchLightning/pytorch-lightning/pull/4648))
 
 
+- Changed `automatic_optimization` to be a model attribute ([#4602](https://github.com/PyTorchLightning/pytorch-lightning/pull/4602))
+
+
+- Changed `Simple Profiler` report to order by percentage time spent + num calls ([#4880](https://github.com/PyTorchLightning/pytorch-lightning/pull/4880))
+
+
 ### Deprecated
 
 - Deprecated `prefix` argument in `ModelCheckpoint` ([#4765](https://github.com/PyTorchLightning/pytorch-lightning/pull/4765))
 
 
+- Deprecated the old way of assigning hyper-parameters through `self.hparams = ...` ([#4813](https://github.com/PyTorchLightning/pytorch-lightning/pull/4813))
+
+
+- Deprecated `mode='auto'` from `ModelCheckpoint` and `EarlyStopping` ([#4695](https://github.com/PyTorchLightning/pytorch-lightning/pull/4695))
+
+
 ### Removed
 
 
 
 ### Fixed
 
+- Added feature to move tensors to CPU before saving ([#4309](https://github.com/PyTorchLightning/pytorch-lightning/pull/4309))
+
+- Fixed `LoggerConnector` to have logged metrics on root device in DP ([#4138](https://github.com/PyTorchLightning/pytorch-lightning/pull/4138))
+
+
+- Auto convert tensors to contiguous format when `gather_all` ([#4907](https://github.com/PyTorchLightning/pytorch-lightning/pull/4907))
 
 
 ## [1.0.8] - 2020-11-24
@@ -82,6 +138,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Renamed class metric `Fbeta` >> `FBeta` ([#4656](https://github.com/PyTorchLightning/pytorch-lightning/pull/4656))
 - Model summary: add 1 decimal place ([#4745](https://github.com/PyTorchLightning/pytorch-lightning/pull/4745))
 - Do not override `PYTHONWARNINGS` ([#4700](https://github.com/PyTorchLightning/pytorch-lightning/pull/4700))
+- Changed `init_ddp_connection` moved from `DDP` to `DDPPlugin` ([#4407](https://github.com/PyTorchLightning/pytorch-lightning/pull/4407))
+
 
 ### Fixed
 
@@ -122,6 +180,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Added `manual_optimizer_step` which work with `AMP Native` and `accumulated_grad_batches` ([#4485](https://github.com/PyTorchLightning/pytorch-lightning/pull/4485))
 - Added `persistent(mode)` method to metrics, to enable and disable metric states being added to `state_dict` ([#4482](https://github.com/PyTorchLightning/pytorch-lightning/pull/4482))
 - Added congratulations at the end of our notebooks ([#4555](https://github.com/PyTorchLightning/pytorch-lightning/pull/4555))
+- Added parameters `move_metrics_to_cpu` in Trainer to disable gpu leak ([#4592](https://github.com/PyTorchLightning/pytorch-lightning/pull/4592))
+
 
 ### Changed
 
@@ -141,7 +201,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Replace `MisconfigurationException` with warning in `ModelCheckpoint` Callback ([#4560](https://github.com/PyTorchLightning/pytorch-lightning/pull/4560))
 - Fixed logged keys in mlflow logger ([#4412](https://github.com/PyTorchLightning/pytorch-lightning/pull/4412))
 - Fixed `is_picklable` by catching `AttributeError` ([#4508](https://github.com/PyTorchLightning/pytorch-lightning/pull/4508))
-
+- Fixed multi test dataloaders dict `AttributeError` error ([#4480](https://github.com/PyTorchLightning/pytorch-lightning/pull/4480))
+- Fixed show progress bar only for `progress_rank 0` on `DDP_SLURM` ([#4437](https://github.com/PyTorchLightning/pytorch-lightning/pull/4437))
 
 ## [1.0.5] - 2020-11-03
 
@@ -156,6 +217,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Hook `on_after_backward` is called only when `optimizer_step` is being called ([#4439](https://github.com/PyTorchLightning/pytorch-lightning/pull/4439))
 - Moved `track_and_norm_grad` into `training loop` and called only when `optimizer_step` is being called ([#4439](https://github.com/PyTorchLightning/pytorch-lightning/pull/4439))
 - Changed type checker with explicit cast of `ref_model` object ([#4457](https://github.com/PyTorchLightning/pytorch-lightning/pull/4457))
+- Changed `distributed_backend` -> `accelerator` ([#4429](https://github.com/PyTorchLightning/pytorch-lightning/pull/4429))
 
 ### Deprecated
 
@@ -172,6 +234,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed TorchScript trace method's data to device and docstring ([#4360](https://github.com/PyTorchLightning/pytorch-lightning/pull/4360))
 - Fixed CSV logger warning ([#4419](https://github.com/PyTorchLightning/pytorch-lightning/pull/4419))
 - Fixed skip DDP parameter sync ([#4301](https://github.com/PyTorchLightning/pytorch-lightning/pull/4301))
+- Fixed `WandbLogger` _sanitize_callable function ([#4422](https://github.com/PyTorchLightning/pytorch-lightning/pull/4422))
+- Fixed `AMP Native` `_unscale` gradient ([#4441](https://github.com/PyTorchLightning/pytorch-lightning/pull/4441))
+
 
 ## [1.0.4] - 2020-10-27
 
@@ -183,6 +248,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Added `fsspec` support for profilers ([#4162](https://github.com/PyTorchLightning/pytorch-lightning/pull/4162))
 - Added autogenerated helptext to `Trainer.add_argparse_args` ([#4344](https://github.com/PyTorchLightning/pytorch-lightning/pull/4344))
 - Added support for string values in `Trainer`'s `profiler` parameter ([#3656](https://github.com/PyTorchLightning/pytorch-lightning/pull/3656))
+- Added support for string values in `Trainer`'s `profiler` parameter ([#3656](https://github.com/PyTorchLightning/pytorch-lightning/pull/3656))
+- Added `optimizer_closure` to `optimizer.step` when supported ([#4190](https://github.com/PyTorchLightning/pytorch-lightning/pull/4190))
+- Added unification of regression metrics ([#4166](https://github.com/PyTorchLightning/pytorch-lightning/pull/4166))
+- Added checkpoint load from Bytes ([#4314](https://github.com/PyTorchLightning/pytorch-lightning/pull/4314))
 
 ### Changed
 
@@ -202,6 +271,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed setting device ids in DDP ([#4297](https://github.com/PyTorchLightning/pytorch-lightning/pull/4297))
 - Fixed synchronization of best model path in `ddp_accelerator` ([#4323](https://github.com/PyTorchLightning/pytorch-lightning/pull/4323))
 - Fixed `WandbLogger` not uploading checkpoint artifacts at the end of training ([#4341](https://github.com/PyTorchLightning/pytorch-lightning/pull/4341))
+- Fixed `FBeta` computation ([#4183](https://github.com/PyTorchLightning/pytorch-lightning/pull/4183))
+- Fixed `accumulation across batches` has completed `before breaking training loop` ([#4278](https://github.com/PyTorchLightning/pytorch-lightning/pull/4278))
+- Fixed `ModelCheckpoint` don't increase current_epoch and global_step when not training ([#4291](https://github.com/PyTorchLightning/pytorch-lightning/pull/4291))
+- Fixed `COMET_EXPERIMENT_KEY` environment variable usage in comet logger ([#4230](https://github.com/PyTorchLightning/pytorch-lightning/pull/4230))
 
 ## [1.0.3] - 2020-10-20
 

MANIFEST.in

@@ -42,6 +42,7 @@ exclude tests
 recursive-exclude docs *
 exclude docs
 recursive-include docs/source/_images/logos/ *
+recursive-include docs/source/_images/badges/ *
 recursive-include docs/source/_images/general/ pl_overview* tf_* tutorial_* PTL101_*
 
 # Include the Requirements