Merge branch 'master' into rename/accelerator_type

Author: kaushikb11

.github/workflows/code-checks.yml

@@ -14,8 +14,8 @@ jobs:
       - uses: actions/setup-python@v2
         with:
           python-version: 3.9
-      - name: Install mypy
+      - name: Install dependencies
         run: |
-          grep mypy requirements/test.txt | xargs -0 pip install
+          pip install '.[dev]'
           pip list
       - run: mypy --install-types --non-interactive

CHANGELOG.md

@@ -13,20 +13,26 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 
 - Fault Tolerant Manual
+    * Add `_SupportsStateDict` protocol to detect if classes are stateful ([#10646](https://github.com/PyTorchLightning/pytorch-lightning/issues/10646))
     * Add `_FaultTolerantMode` enum used to track different supported fault tolerant modes ([#10645](https://github.com/PyTorchLightning/pytorch-lightning/issues/10645))
+    * Add a `_rotate_worker_indices` utility to reload the state according the latest worker ([#10647](https://github.com/PyTorchLightning/pytorch-lightning/issues/10647))
+    * Add stateful workers ([#10674](https://github.com/PyTorchLightning/pytorch-lightning/issues/10674))
+    * Add an utility to collect the states across processes ([#10639](https://github.com/PyTorchLightning/pytorch-lightning/issues/10639))
+    * Add logic to reload the states across data loading components ([#10699](https://github.com/PyTorchLightning/pytorch-lightning/issues/10699))
+    * Cleanup some fault tolerant utilities ([#10703](https://github.com/PyTorchLightning/pytorch-lightning/issues/10703))
+    * Enable Fault Tolerant Manual Training ([#10707](https://github.com/PyTorchLightning/pytorch-lightning/issues/10707))
 
-
--
+- Added support for re-instantiation of custom (subclasses of) `DataLoaders` returned in the `*_dataloader()` methods, i.e., automatic replacement of samplers now works with custom types of `DataLoader` ([#10680](https://github.com/PyTorchLightning/pytorch-lightning/issues/10639))
 
 
--
+- Show a better error message when a custom `DataLoader` implementation is not well implemented and we need to reconstruct it ([#10719](https://github.com/PyTorchLightning/pytorch-lightning/issues/10719))
 
 
 -
 
 ### Changed
 
-- Raise exception in `init_dist_connection()` when torch distibuted is not available ([#10418](https://github.com/PyTorchLightning/pytorch-lightning/issues/10418))
+- Raised exception in `init_dist_connection()` when torch distibuted is not available ([#10418](https://github.com/PyTorchLightning/pytorch-lightning/issues/10418))
 
 
 - The `monitor` argument in the `EarlyStopping` callback is no longer optional ([#10328](https://github.com/PyTorchLightning/pytorch-lightning/pull/10328))
@@ -35,7 +41,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Do not fail if batch size could not be inferred for logging when using DeepSpeed ([#10438](https://github.com/PyTorchLightning/pytorch-lightning/issues/10438))
 
 
-- Raise `MisconfigurationException` when `enable_progress_bar=False` and a progress bar instance has been passed in the callback list ([#10520](https://github.com/PyTorchLightning/pytorch-lightning/issues/10520))
+- Raised `MisconfigurationException` when `enable_progress_bar=False` and a progress bar instance has been passed in the callback list ([#10520](https://github.com/PyTorchLightning/pytorch-lightning/issues/10520))
 
 
 - Moved `trainer.connectors.env_vars_connector._defaults_from_env_vars` to `utilities.argsparse._defaults_from_env_vars` ([#10501](https://github.com/PyTorchLightning/pytorch-lightning/pull/10501))
@@ -50,6 +56,12 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Moved ownership of the `PrecisionPlugin` into `TrainingTypePlugin` and updated all references ([#10570](https://github.com/PyTorchLightning/pytorch-lightning/pull/10570))
 
 
+- Raised an error if the `batch_size` cannot be inferred from the current batch if it contained a string or was a custom batch object ([#10541](https://github.com/PyTorchLightning/pytorch-lightning/pull/10541))
+
+
+-
+
+
 -
 
 
@@ -69,6 +81,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `DeviceType` in favor of `_AcceleratorType` ([#10503](https://github.com/PyTorchLightning/pytorch-lightning/pull/10503))
 
 
+- Deprecated the property `Trainer.slurm_job_id` in favor of the new `SLURMEnvironment.job_id()` method ([#10622](https://github.com/PyTorchLightning/pytorch-lightning/pull/10622))
+
+
 -
 
 ### Removed
@@ -154,6 +169,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Removed the `precision_plugin` attribute from `Accelerator` in favor of its equivalent attribute `precision_plugin` in the `TrainingTypePlugin` ([#10570](https://github.com/PyTorchLightning/pytorch-lightning/pull/10570))
 
+
+- Removed `DeepSpeedPlugin.{precision,amp_type,amp_level}` properties ([#10657](https://github.com/PyTorchLightning/pytorch-lightning/pull/10657))
+
+
 ### Fixed
 
 - When a tensor is logged with `self.log`, run its computation with the same `dtype` ([#10076](https://github.com/PyTorchLightning/pytorch-lightning/pull/10076))
@@ -171,10 +190,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed an issue that caused Lightning to extract the batch size even though it was set by the user in `LightningModule.log` ([#10408](https://github.com/PyTorchLightning/pytorch-lightning/pull/10408))
 
 
--
+- Fixed `Trainer(move_metrics_to_cpu=True)` not moving the evaluation logged results to CPU ([#10631](https://github.com/PyTorchLightning/pytorch-lightning/pull/10631))
 
 
--
+- Fixed the `{validation,test}_step` outputs getting moved to CPU with `Trainer(move_metrics_to_cpu=True)` ([#10631](https://github.com/PyTorchLightning/pytorch-lightning/pull/10631))