Label tracking meta-issue (edit me to get automatically CC'ed on issues!) #10530
Description
This issue is used by lightning-probot to manage subscriptions to labels. To subscribe yourself to a label, add a line * label @yourusername, or add your username to an existing line (space separated) in the body of this issue. Do not try to subscribe in comments, the bot only parses the initial post.
This is a copy of pytorch/pytorch#24422.
As a courtesy to others, please do not edit the subscriptions of users who are not you.
Note: Some labels have sublabels (e.g. callback), but you won't be subscribing to them automatically if you choose the label at the top level.
PR status
- ready @lantiga
- has conflicts
- won’t fix
Issue/PR category
- design @lantiga @tchaton @justusschock
- breaking change @lantiga @justusschock
- bug @ethanwharris
- code quality
- deprecation @lantiga
- discussion @lantiga
- docs @lantiga
- feature @lantiga
- question @lantiga
- refactor @lantiga @justusschock
- release @lantiga
Issue/PR severity
Issue/PR metadata
- duplicate @ethanwharris
- good first issue @ethanwharris
- help wanted
- admin @ethanwharris
- let’s do it! @tchaton
- waiting on author
- working as intended
- 3rd party
- experimental @ethanwharris
- needs triage
- community
Generic labels
- pl
Code section (pl)
- accelerator @justusschock @lantiga
- accelerator: cpu @justusschock @lantiga
- accelerator: cuda @justusschock @lantiga
- accelerator: hpu (external) @jerome-habana
- accelerator: mps @justusschock @lantiga
- accelerator: tpu @JackCaoG @Liyang90 @gkroiz
- callback @lantiga
- callback: device stats
- callback: early stopping
- callback: fine-tuning
- callback: gradient accumulation
- callback: lambda function
- callback: lr monitor
- callback: model checkpoint
- callback: model summary
- callback: prediction writer
- callback: pruning
- callback: swa
- callback: timer
- callback: throughput
- ci @ethanwharris
- environment @lantiga
- environment: kubeflow @lantiga
- environment: lightning @lantiga
- environment: lsf @lantiga
- environment: slurm @lantiga
- environment: torchelastic @lantiga
- environment: mpi @lantiga
- hooks @lantiga @justusschock
- io @justusschock
- lightningcli @mauvilsa
- lightningdatamodule @lantiga
- fabric @justusschock @lantiga
- lightningmodule @justusschock @lantiga
- logger @lantiga
- logger: comet
- logger: csv @ethanwharris
- logger: mlflow
- logger: neptune
- logger: tensorboard @lantiga
- logger: wandb @lantiga @morganmcg1 @borisdayma @scottire @parambharat
- logging
- loops @justusschock
- lr scheduler
- optimizer
- plugin @justusschock @lantiga
- precision: double @justusschock @lantiga
- precision: amp @justusschock @lantiga
- precision: bnb @lantiga
- precision: te @lantiga
- precision: half @lantiga
- profiler
- progress bar: rich
- progress bar: tqdm @lantiga
- progress tracking (internal) @lantiga
- strategy @justusschock @lantiga
- strategy: ddp @justusschock @lantiga
- strategy: deepspeed @lantiga
- strategy: dp (removed in pl) @justusschock @lantiga
- strategy: fsdp @lantiga
- strategy: hpu (external) @jerome-habana
- strategy: hivemind (external)
- strategy: xla @JackCaoG @Liyang90 @gkroiz
- tests @lantiga
- trainer @lantiga @justusschock
- trainer: connector @justusschock @lantiga
- trainer: fit
- trainer: argument @justusschock @lantiga
- trainer: predict
- trainer: test
- trainer: tune
- trainer: validate
- torch.compile
- tuner
Feature (pl)
- fault tolerance @justusschock
- checkpointing @lantiga
- data handling @tchaton
- distributed @justusschock
- optimization @ethanwharris
- reproducibility @lantiga
- performance @lantiga