Merge branch 'master' into deprecate-write-predictions

.github/workflows/ci_test-full.yml

@@ -6,11 +6,14 @@ on:  # Trigger the workflow on push or pull request, but only for the master bra
     branches: [master, "release/*"]
   pull_request:
     branches: [master, "release/*"]
+    types: [opened, reopened, ready_for_review, synchronize]
 
 jobs:
+
   pytest:
 
     runs-on: ${{ matrix.os }}
+    if: github.event.pull_request.draft == false
     strategy:
       fail-fast: false
       matrix:

.github/workflows/ci_test-tpu.yml

@@ -23,7 +23,7 @@ jobs:
       fail-fast: false
       matrix:
         python-version: [3.7]
-        xla-version: [1.6, 1.7]
+        xla-version: [1.6, 1.8]
     # Timeout: https://stackoverflow.com/a/59076067/4521646
     timeout-minutes: 50
 

CHANGELOG.md

@@ -102,6 +102,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Added `max_time` Trainer argument to limit training time ([#6823](https://github.com/PyTorchLightning/pytorch-lightning/pull/6823))
 
 
+- Added new `EarlyStopping` parameters `stopping_threshold` and `divergence_threshold` ([#6868](https://github.com/PyTorchLightning/pytorch-lightning/pull/6868))
+
+
+
 ### Changed
 
 - Renamed `pytorch_lightning.callbacks.swa` to `pytorch_lightning.callbacks.stochastic_weight_avg` ([#6259](https://github.com/PyTorchLightning/pytorch-lightning/pull/6259))
@@ -128,6 +132,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Changed warnings and recommendations for dataloaders in `ddp_spawn` ([#6762](https://github.com/PyTorchLightning/pytorch-lightning/pull/6762/))
 
 
+- `pl.seed_everyting` will now also set the seed on the `DistributedSampler` ([#7024](https://github.com/PyTorchLightning/pytorch-lightning/pull/7024))
+
+
 ### Deprecated
 
 - Deprecated `LightningModule.write_predictions` and `LigtningModule.write_predictions_dict` ([#7066](https://github.com/PyTorchLightning/pytorch-lightning/pull/7066))
@@ -148,6 +155,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `PytorchProfiler(profiled_functions)` in favor of `record_functions` ([#6349](https://github.com/PyTorchLightning/pytorch-lightning/pull/6349))
 
 
+- Deprecated `@auto_move_data` in favor of `trainer.predict` ([#6993](https://github.com/PyTorchLightning/pytorch-lightning/pull/6993))
+
+
 - Deprecated metrics in favor of `torchmetrics` ([#6505](https://github.com/PyTorchLightning/pytorch-lightning/pull/6505),
     [#6530](https://github.com/PyTorchLightning/pytorch-lightning/pull/6530),
     [#6540](https://github.com/PyTorchLightning/pytorch-lightning/pull/6540),
@@ -288,6 +298,12 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed process rank not being available right away after `Trainer` instantiation ([#6941](https://github.com/PyTorchLightning/pytorch-lightning/pull/6941))
 
 
+- Fixed the order to call for world ranks & the `root_device` property in `TPUSpawnPlugin` ([#7074](https://github.com/PyTorchLightning/pytorch-lightning/pull/7074))
+
+
+- Fixed metric objects passed directly to `self.log` not being reset correctly ([#7055](https://github.com/PyTorchLightning/pytorch-lightning/pull/7055))
+
+
 ## [1.2.7] - 2021-04-06
 
 ### Fixed

docs/source/advanced/multi_gpu.rst

@@ -102,6 +102,8 @@ Lightning adds the correct samplers when needed, so no need to explicitly add sa
 .. note::
     By default it will add ``shuffle=True`` for train sampler and ``shuffle=False`` for val/test sampler.
     ``drop_last`` in :class:`~torch.utils.data.distributed.DistributedSampler` will be set to its default value in PyTorch.
+    If you called :func:`~pytorch_lightning.utilities.seed.seed_everyting`, Lightning will set the same seed for the
+    sampler.
 
 .. note:: You can disable this behavior with ``Trainer(replace_sampler_ddp=False)``
 

docs/source/common/lightning_module.rst

@@ -698,6 +698,12 @@ log_dict
 .. automethod:: pytorch_lightning.core.lightning.LightningModule.log_dict
     :noindex:
 
+manual_backward
+~~~~~~~~~~~~~~~
+
+.. automethod:: pytorch_lightning.core.lightning.LightningModule.manual_backward
+    :noindex:
+
 print
 ~~~~~
 
@@ -916,7 +922,10 @@ True if using Automatic Mixed Precision (AMP)
 
 automatic_optimization
 ~~~~~~~~~~~~~~~~~~~~~~
-When set to ``False``, Lightning does not automate the optimization process. This means you are responsible for handling your optimizers. However, we do take care of precision and any accelerators used.
+When set to ``False``, Lightning does not automate the optimization process. This means you are responsible for handling
+your optimizers. However, we do take care of precision and any accelerators used.
+
+See :ref:`manual optimization<common/optimizers:Manual optimization>` for details.
 
 .. code-block:: python
 
@@ -931,7 +940,9 @@ When set to ``False``, Lightning does not automate the optimization process. Thi
         self.manual_backward(loss)
         opt.step()
 
-This is recommended only if using 2+ optimizers AND if you know how to perform the optimization procedure properly. Note that automatic optimization can still be used with multiple optimizers by relying on the ``optimizer_idx`` parameter. Manual optimization is most useful for research topics like reinforcement learning, sparse coding, and GAN research.
+This is recommended only if using 2+ optimizers AND if you know how to perform the optimization procedure properly. Note
+that automatic optimization can still be used with multiple optimizers by relying on the ``optimizer_idx`` parameter.
+Manual optimization is most useful for research topics like reinforcement learning, sparse coding, and GAN research.
 
 .. code-block:: python
 
@@ -1086,13 +1097,6 @@ get_progress_bar_dict
 .. automethod:: pytorch_lightning.core.lightning.LightningModule.get_progress_bar_dict
     :noindex:
 
-manual_backward
-~~~~~~~~~~~~~~~
-
-.. automethod:: pytorch_lightning.core.lightning.LightningModule.manual_backward
-    :noindex:
-
-
 on_after_backward
 ~~~~~~~~~~~~~~~~~