ray-project/ray_lightning

Issues

• Deprecation Notice: ray_lightning to be Replaced with New LightningTrainer in Ray 2.4

  #258 opened 2 years ago by woshiyyya
  1

• Trails did not complete error

  #257 opened 2 years ago by Bk073
  0

• ray-lightning doesn't seem to be compatible with ray 2.3

  #255 opened 2 years ago by marcosrdac
  2

• Trials hang when using a scheduler

  #253 opened 2 years ago by dcfidalgo
  0

• What happens with custom samplers?

  #252 opened 2 years ago by AugustoPeres
  0

• RuntimeError: Error(s) in loading state_dict: Unexpected key(s) when recovering results from main process during Trainer.fit()

  #246 opened 2 years ago by davzaman
  0

• TPU support?

  #245 opened 2 years ago by platers
  0

• [Question] Is it necessary to adapt report and Checkpointing to the newly introduced session and Checkpoint API of Ray AIR?

  #243 opened 2 years ago by MarkusSpanring
  0

• Error when using WandbLogger

  #205 opened 2 years ago by KwanWaiChung
  1

• Multi-GPU training fails with `ValueError` on systems with UUID GPU IDs

  #236 opened 2 years ago by davidegraff
  1

• TypeError in a SLURM environment due to internal API break

  #235 opened 2 years ago by dcfidalgo
  2

• [Code] add pytorch-lightning compatibility for 1.7.x

  #194 opened 2 years ago by JiahaoYao
  6

• Rank Zero Deprecation

  #230 opened 2 years ago by lcaquot94
  1

• Can not checkpoint and log

  #228 opened 2 years ago by lcaquot94
  1

• Ray lightning opens a new mlflow run

  #225 opened 2 years ago by AugustoPeres
  0

• TuneReportCheckpointCallback error

  #219 opened 2 years ago by jakubMitura14
  2

• population based scheduler error

  #220 opened 2 years ago by jakubMitura14
  0

• no training starts although flag is running

  #216 opened 2 years ago by jakubMitura14
  5

• Teardown after trainer.fit() takes exceptionally long when using RayStrategy with large models

  #207 opened 2 years ago by MarkusSpanring
  1

• Deterministic mode is not set on remote worker when `Trainer` is set to `deterministic`

  #213 opened 2 years ago by MarkusSpanring
  0

• Question: Why use ray_lightning instead of pytorch_lightning for multi-node training?

  #212 opened 2 years ago by saryazdi
  4

• Worker nodes don't start for ray-lightning & aws

  #210 opened 2 years ago by toru34
  1

• [Code] best_model_path in ModelCheckpointCallback (rank 0 and driver node)

  #202 opened 2 years ago by chongxiaoc
  0

• Error in RayStrategy.root_device when using multi GPU node

  #192 opened 2 years ago by m-lyon
  20

• adding the version in `__init__`

  #191 opened 2 years ago by JiahaoYao
  0

• AttributeError: 'AcceleratorConnector' object has no attribute 'strategy'

  #189 opened 2 years ago by m-lyon
  6

• Distributed training performance slowdown when resuming from a checkpoint.

  #184 opened 2 years ago by subhashbylaiah
  5

• `ray_lightning` checkpoint dir not saving the checkpoint

  #186 opened 2 years ago by JiahaoYao
  0

• `ray_horovod` leaks gpu memory on the `cuda:0`

  #181 opened 2 years ago by JiahaoYao
  2

• `ray_horovod` multi pid process in the `run`

  #182 opened 2 years ago by JiahaoYao
  2

• `ray_ddp` issue of `Leaking Caffe2 thread-pool after fork. (function pthreadpool)`

  #180 opened 2 years ago by JiahaoYao
  0

• `ray_ddp` gpu issue

  #179 opened 2 years ago by JiahaoYao
  3

• `ray_ddp` global and local rank

  #175 opened 2 years ago by JiahaoYao
  1

• tune test: do we need to count the head node cpu?

  #178 opened 2 years ago by JiahaoYao
  1

• `ray_ddp` showing no use of gpu

  #177 opened 2 years ago by JiahaoYao
  2

• `ray_ddp` the progressive bar is broken

  #176 opened 2 years ago by JiahaoYao
  1

• ray ddp fails with 2 gpu workers

  #174 opened 2 years ago by JiahaoYao
  10

• `shard-ddp` test of system exit

  #173 opened 2 years ago by JiahaoYao
  0

• warning in the ci test (change the deprecated api)

  #172 opened 2 years ago by JiahaoYao
  1

• torch remove the checkpoint when `is_global_zero` is not set? (multi-worker setting)

  #171 opened 2 years ago by JiahaoYao
  0

• log is changed in the new version of pytorch lightning

  #170 opened 2 years ago by JiahaoYao
  1

• change the `checkpoint_callback=True`

  #169 opened 2 years ago by JiahaoYao
  0

• warning from the horovod trainer

  #168 opened 2 years ago by JiahaoYao
  3

• horovod lightning integration missing the log dir

  #167 opened 2 years ago by JiahaoYao
  0

• horovod installation issue

  #165 opened 2 years ago by JiahaoYao
  1

• trainer is not consistent during the `ray_ddp`

  #160 opened 2 years ago by JiahaoYao
  1

• the training results can be pulled to the main process

  #162 opened 2 years ago by JiahaoYao
  0

• [raystrategy] multi-stragy in the worker is not consistent

  #161 opened 2 years ago by JiahaoYao
  2

• Using LightningCLI to parse plugin options from the config file fails when using the RayPlugin.

  #151 opened 3 years ago by subhashbylaiah
  1

• Multiple WandB experiments created with PyTorch lightning DDP

  #155 opened 3 years ago by brodyh
  0