1.3.x cherries 🍒

[carmocca]

1.3.1 (#7477)

#7456 6d82dc8 Pin Sphinx<4.0
#7395 6bc616d fix local rank display bug in info message
#7441 1af42d7 Update worker_init_fn test for pytorch 1.9 nightly
#7434 710b144 Restore trainer.current_epoch after tuning
#7414 1a27c12 update ngc for 1.3
#7362 98b94b8 Fix DeepSpeedPlugin with IterableDataset

1.3.2 (#7589)

#7032 20f6337 Fix the condition for calling update_learning_rates
#7492 6ac16ff Fix DistribType for ddp_cpu (spawn)
#7433 7b283e3 Bugfix/Multiple dataloaders
#7238 b65ae79 Automatically check DataModule.has_{setup,teardown,prepare_data} [2/2]

1.3.3 (#7712)

#7692 8ba6304 Increment the total batch idx before the accumulation early exit
#7683 2c10ecc MAINTAINER has been deprecated
#7677 8b01497 Fix global step update when the epoch is skipped
#7415 3f460b1 Move parameter validation specific to TPU Training plugins
#7674 a54bc5d Fix progress bar print error when called before training
#7566 0c958c5 Fix dataloaders are not reset when tuning the model
#7563 01109cd Fix/mismatched toggle optimizer
#7592 92cf396 Override broadcast_object_list for torch<1.8
#7608 ed27190 Clear predict_progress_bar in ProgressBar.__getstate__

1.3.4 (#7786)

#7781 0b6fd1d Update pre-commit and add new hooks
#7780 7e6010f fix info message when max training time reached
#7681 04dcb17 Add __len__ method to IndexBatchSamplerWrapper

1.3.5 (#7881)

#7868 f9fccdf Move training_output validation to after train_step_end
#7869 ce97676 update fsspec to 2021.06.0
#7835 d1becce [bugfix] Resolve LearningRateMonitor + BackboneFinetuning
#7851 009e05d [bugfix] Minor improvements to apply_to_collection and type signature of log_dict
#7834 7b531ac fix docker versions
#7779 6a0d503 Add warning to trainstep output
#7825 ca89a7f [sharded plugin] Fix check for fp16 precision
#7807 f345840 Fix support for torch Module type hints in LightningCLI

1.3.6 (#7986)

#7993 92024df Pt 1.9 breaking fix: iter type hint
#7931 3a0ed02 Properly handle parent modules w/ parameters in BaseFinetuning callback
#7958 cca0e75 remove parsing comments
#7942 22d8266 Seed all workers when using DDP
#7967 20a5e09 fix myst-parser warning blocking docs ci
#7945 6856cce Remove rank_zero_only on DataModule prepare_data
#7889 7f4ef6d Fix logs overwriting issue for remote fs

gh pr list -s merged -S 'merged:2021-06-08T16:30:00.000Z..2021-06-15T12:00:00.000Z' --json mergedAt,milestone,url,mergeCommit,title --jq 'sort_by(.mergedAt) | reverse | .[] | select((.milestone.title == "v1.3.x") or (.milestone.title == null)) | [.url, .mergeCommit.oid, .title] | join(" ")' --limit 100

Including non-bugfix commits is allowed if they are necessary to support a bugfix commit and it does not break compatibility. In that case, please add a comment on why the commit was included.

---

Utils

Update the merged timestamp range as appropriate

Requires gh version >=1.10.2

Commits

The following command provides a list of commits to master (ordered by merge date from newest to oldest) with the appropriate milestone (null or v1.3.x).

gh pr list -s merged -S 'merged:2021-06-01T16:30:00.000Z..2021-06-08T16:30:00.000Z' --json mergedAt,milestone,url,mergeCommit,title --jq 'sort_by(.mergedAt) | reverse | .[] | select((.milestone.title == "v1.3.x") or (.milestone.title == null)) | [.url, .mergeCommit.oid, .title] | join(" ")' --limit 100

Author list

For a bug-fix release

gh pr list -s merged -S 'merged:2021-06-08T16:30:00.000Z..2021-06-15T12:00:00.000Z' --json author,milestone --jq '.[] | select((.milestone.title == "v1.3.x") or (.milestone.title == null)) | ["@" + .author.login] | join("")' --limit 100 | sort --ignore-case | uniq | xargs echo 

For a feature release

gh pr list -s merged --json author,milestone --jq '.[] | select(.milestone.title == "v1.4") | ["@" + .author.login] | join("")' --limit 8000 | sort --ignore-case | uniq | xargs echo

[carmocca]

cc @ethanwharris @PyTorchLightning/core-contributors

[Borda]

@carmocca the script is really GREAT!!

[carmocca]

Thank you @Borda 😊

[tchaton]

@carmocca I second that ! Really neat !