Remove deprecated code after the 1.6 release

[akihironitta]

Proposed refactor

Remove deprecated code after the 1.6 release.

NOTE: Please pick up a single item from the list (by commenting here in the issue) - and if there are no conflicts - we will happily assign you and put your name in front of the item in the list.

Please note that unless mentioned, the classes are importable from pytorch_lightning, example: from pytorch_lightning import Trainer.

• LightningModule.summarize -> remove LightningModule.summarize #12559
• pytorch_lightning.core.memory.LayerSummary -> Remove deprecated LayerSummary and ModelSummary #12593
• pytorch_lightning.core.memory.ModelSummary -> Remove deprecated LayerSummary and ModelSummary #12593
• pytorch_lightning.core.memory.get_gpu_memory_map -> Remove pytorch_lightning.core.memory.get_gpu_memory_map #12644
• pytorch_lightning.core.memory.get_memory_profile -> Remove deprecated get_memory_profile #12659
• LightningModule.model_size -> remove deprecated model_size from LightningModule #12641
• LightningDataModule.train_transforms -> Remove train_transforms in LightningDataModule #12662
• LightningDataModule.val_transforms -> Remove the deprecated LightningDataModule.val_transforms #12763
• LightningDataModule.test_transforms -> Remove the deprecated LightningDataModule.test_transforms  #12773
• LightningDataModule.size -> Remove the deprecated LightningDataModule.size, LightningDataModule.dims #12780
• LightningDataModule.dims and LightningDataModule(dims=...) -> Remove the deprecated LightningDataModule.size, LightningDataModule.dims #12780
• LightningModule.get_progress_bar_dict -> Remove the deprecated get_progress_bar_dict #12839
• Trainer.progress_bar_dict -> Remove the deprecated get_progress_bar_dict #12839
• Trainer(prepare_data_per_node=...) -> Remove deprecated prepare_data_per_node in Trainer #12536
• Trainer(stochastic_weight_avg=...) -> Remove deprecated stochastic_weight_avg argument from Trainer #12535
• Trainer(terminate_on_nan=...) and Trainer.terminate_on_nan -> Remove deprecated 'terminate_on_nan' argument from Trainer #12553
• LightningModule.on_{train,val,test,predict}_dataloader -> Remove the deprecated on_{train,val,test,predict}_dataloader hooks #13033
• pytorch_lightning.loggers.TestTubeLogger -> Remove deprecated TestTubeLogger #12859
• pytorch_lightning.Callback.on_keyboard_interrupt -> Remove deprecated on_keyboard_interrupt #13438
• Trainer(process_position=...) -> Removed process_position argument from Trainer Class #13071
• Trainer(flush_logs_every_n_steps=...) -> Removed flush_logs_every_n_steps argument from Trainer #13074
• LightningModule.add_to_queue -> @shenoynikhil
• LightningModule.get_from_queue -> @shenoynikhil
• Trainer(progress_bar_refresh_rate=...) -> Remove deprecated progress_bar_refresh_rate from Trainer constructor #12514
• LightningLoggerBase.close and pytorch_lightning.loggers.LoggerCollection.close -> Remove the deprecated logger.close #13149
• pytorch_lightning.distributed.dist.LightningDistributed Remove deprecated LightningDistributed #13549
• Trainer(checkpoint_callback=...) -> Remove deprecated checkpoint_callback flag in Trainer #13027
• Passing dataloader_idx to on_train_batch_start of pytorch_lightning.Callback and LightningModule -> Remove deprecated dataloader_idx argument from on_train_batch_start/end callback hooks #12769
• LightningModule.on_post_move_to_device Remove deprecated LightningModule.on_post_move_to_device #13548
• pytorch_lightning.core.decorators.parameter_validation Remove deprecated pytorch_lightning.core.decorators.parameter_validation #13514
• Trainer(accelerator="ddp_spawn") Remove support for passing strategy strings to accelerator #12696
• Trainer(plugins="ddp_spawn") Remove support for passing strategy name to plugins #12700
• Trainer(weights_summary="full"), Trainer(weights_summary=None), Trainer.weights_summary -> Removed deprecated weights_summary argument from Trainer #13070
• Trainer(log_gpu_memory=...) -> Remove deprecated automatic logging of gpu metrics #12657
• Trainer.slurm_job_id Remove deprecated Trainer.slurm_job_id #13459
• pytorch_lightning.callbacks.gpu_stats.GPUStatsMonitor -> Remove deprecated GPUStatsMonitor callback #12554
• pytorch_lightning.callbacks.gpu_stats.XLAStatsMonitor -> Remove deprecated XLAStatsMonitor #12688
• pytorch_lightning.callbacks.progress.ProgressBar -> Remove the deprecated pl.callbacks.ProgressBar #12658
• Trainer(max_steps=None) and Trainer.fit_loop.max_steps = None Remove deprecated max_steps=None #13591
• pytorch_lightning.callbacks.lr_monitor.LearningRateMonitor.lr_sch_names -> Remove pytorch lightning.callbacks.lr monitor.learning rate monitor.lr_sch_names #13353
• KubeflowEnvironment.is_using_kubeflow, LSFEnvironment.is_using_lsf, TorchElasticEnvironment.is_using_torchelastic Remove deprecated ClustertEnvironment methods #13458
• pytorch_lightning.overrides.distributed.IndexBatchSamplerWrapper.batch_indices Removed deprecated pytorch_lightning.overrides.distributed.IndexBatchSamplerWrapper.batch_indices  #13565
• pytorch_lightning.strategies.SingleDeviceStrategy.post_dispatch Remove deprecated Strategy.post_dispatch #13461
• pytorch_lightning.trainer.connectors.logger_connector.logger_connector.LoggerConnector.gpu_metrics

Feel free to cross-check from the test file to ensure that the relevant test fails now (since it's no more deprecated and instead removed).

Pitch

All the deprecated features we have are tested here:

https://github.com/PyTorchLightning/pytorch-lightning/blob/master/tests/deprecated_api/test_remove_1-7.py

If you are interested in taking care of one item, post a comment here asking to take it. This avoids multiple people working on the same thing.

Additional context

See pull requests linked in #10312 for examples on how to contribute :)
Or a recent pull request #12514.

---

If you enjoy Lightning, check out our other projects! ⚡

• Metrics: Machine learning metrics for distributed, scalable PyTorch applications.

• Lite: enables pure PyTorch users to scale their existing code on any kind of device while retaining full control over their own loops and optimization logic.

• Flash: The fastest way to get a Lightning baseline! A collection of tasks for fast prototyping, baselining, fine-tuning, and solving problems with deep learning.

• Bolts: Pretrained SOTA Deep Learning models, callbacks, and more for research and production with PyTorch Lightning and PyTorch.

• Lightning Transformers: Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

cc @Borda @justusschock @awaelchli @rohitgr7 @krshrimali

[krshrimali]

@akihironitta - Thanks for creating this issue. Let's create a tracker here, to help track better? What do you think?

[Borda]

Thanks for creating this issue. Let's create a tracker here, to help track better? What do you think?

exactly, let's include in the issue list aka todo of heat methods/attributes shall be removed so individual contributors can pick one and we reduce collision of people working on the same... 🐰

[krshrimali]

@Borda - Done! 🚀 Note and a request to maintainers: please cross-check this list.

[yhl48]

I will take Trainer(prepare_data_per_node=...) for a start.

[krshrimali]

I will take the following for a start:

Trainer.progress_bar_dict Trainer(prepare_data_per_node=...) Trainer(stochastic_weight_avg=...) Trainer(terminate_on_nan=...) and Trainer.terminate_on_nan Trainer(process_position=...) Trainer(flush_logs_every_n_steps=...)

Hi, @yhl48 - thank you for showing interest in contributing to PyTorch Lightning. Happy to see the list eventually being completed, however - I've a minor request and a suggestion - can you please pick up one of these as a start, once you are done with that PR - you can comment here to pick up another to work on? This will allow other contributors to also pick up from the list while we all are at work :)) Apologies that we didn't mention it at first in the description, but I'll do it now. 🚀 Looking forward to your PR(s) ⚡

Let me know which one would you like to work on from the list, and I can put your name there in the list :)

[yhl48]

@krshrimali thanks for the suggestion, I have edited the comment to only include one flag.

[krshrimali]

@krshrimali thanks for the suggestion, I have edited the comment to only include one flag.

Assigned to you. Thank you for understanding :) Please let us know if you have any queries, you can also join our slack if you wish: https://join.slack.com/t/pytorch-lightning/shared_invite/zt-12iz3cds1-uyyyBYJLiaL2bqVmMN7n~A. :)

[aniketmaurya]

Hi, I would like to take LightningModule.summarize

[karthikrangasai]

Hello,
I can start with Trainer(terminate_on_nan=...) and Trainer.terminate_on_nan

[kaushikb11]

Note to the contributors here:

It's a good practice to check if any open source projects are using the API, you are planning to remove on grep.app. So, that we could notify them to use an alternative or update the code. :)

For instances: https://grep.app/search?q=GPUStatsMonitor, https://grep.app/search?q=progress_bar_refresh_rate%3D

[aniketmaurya]

Please assign me pytorch_lightning.core.memory.LayerSummary and ModelSummary

[puhuk]

Can I take pytorch_lightning.core.memory.get_gpu_memory_map

[rohitgr7]

sure, go ahead @puhuk !!!

[ankitaS11]

Hi, can I take pytorch_lightning.callbacks.gpu_stats.XLAStatsMonitor ?

[rschireman]

I'll take pytorch_lightning.callbacks.progress.ProgressBar if that's ok

[tkonopka]

Hello, I can work on LightningModule.model_size (?)

[krshrimali]

@ankitaS11 @rschireman @tkonopka - Thank you for showing interest! 🚀 ⚡ Please go ahead, I've assigned the respective tasks to you all (updated the description). Also, make sure that you join our slack channel - feel free to ask questions there if you have any!

[rschireman]

Good morning, can I take Trainer.slurm_job_id?

[rohitgr7]

yep @rschireman go ahead. I see a few more assigned to you. Are you working on them too?

[rschireman]

@rohitgr7 pytorch_lightning.callbacks.lr_monitor.LearningRateMonitor.lr_sch_names was closed yesterday I believe. I don't see any other assignments on my end but I'll gladly close those before I work on this.

[rohitgr7]

oh, I didn't see it since it wasn't marked. Go ahead with the next one :)

[akihironitta]

@shenoynikhil @nitinramvelraj Thanks for showing your interest in contributing! Let me know if you need help with the ones assigned to you :]

[nitinramvelraj]

@akihironitta Hi thanks! I made the changes, I'm having trouble running the test cases. Could you help me with that?

[akihironitta]

Hi @nitinramvelraj Thanks for asking! Let's discuss your issue via PyTorch Lightning Slack! My username there is the same as my GitHub username @akihironitta. Please DM me on Slack :) (You can join our Slack via https://pytorchlightning.ai/community if you haven't)

[nitinramvelraj]

Thanks! I don't have an email id with the following domains:
berkeley.edu
uchicago.edu
ncsu.edu
bcs.tu-darmstadt.de
di.uniroma1.it
maiorca.xyz

Without this the slack is not letting me join. Can one of the administrators give me an invitation to join the Slack?

[akihironitta]

Hi @nitinramvelraj! Sorry for the delay. Could you try this link?
https://join.slack.com/t/pytorch-lightning/shared_invite/zt-1b6lkirxt-9b3ehI0P_31YIqQ4_Dum1A

(The link in the https://pytorchlightning.ai/community will be updated to this new link shortly.)

[shantam-8]

Hello @akihironitta, can I please take pytorch_lightning.core.decorators.parameter_validation?

[rohitgr7]

hey @shantam-8 ! please go ahead with a PR !!

[akihironitta]

Hi @shenoynikhil @madaanpulkit! Thank you for your interest in contributing! Let me know if you need help with the ones assigned to you :)

---

EDIT: @madaanpulkit I'll work on the item assigned to you now. If you're still interested in contributing, please choose one from the good first issues!

[shenoynikhil]

@akihironitta Have been busy for sometime, will work on it today :).

[samz5320]

Hi ! can i take pytorch_lightning.overrides.distributed.IndexBatchSamplerWrapper.batch_indices ? if its not assigned .

[akihironitta]

Hi @samz5320, thank you for showing your interest! Please go ahead :)

[carmocca]

Closing. All the items are taken (some yet to be merged). Thank you all for your help!