open-mmlab/mmengine

Issues

• FSDPStrategy how to set mixed_precision and other params of pytorch

  #1541 opened 15 days ago by apachemycat
  0

• [Bug] 分布式训练代码例子报错，

  #1540 opened 15 days ago by apachemycat
  0

• [Feature] Support calculating loss in the validation step

  #1486 opened 4 months ago by zhouzaida
  1

• ValueError: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. T

  #1539 opened 19 days ago by apachemycat
  0

• [Bug] 中断后恢复训练报错RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

  #1538 opened 22 days ago by Helen-Ren-yi
  0

• [Bug] 多卡情况下，训练后eval和离线test的精度不能保证一致

  #1536 opened 24 days ago by whlook
  0

• [Bug] No module named 'mmengine.models'

  #1525 opened 2 months ago by hitbuyi
  2

• DeepSpeed2 不能自动排除冻结的参数[Bug]

  #1518 opened 24 days ago by Baboom-l
  1

• [Bug] Unable to save results using pklfile_prefix tag

  #1533 opened a month ago by abadithela
  1

• [Feature] Support dataset streaming?

  #1535 opened a month ago by Ablustrund
  0

• [Feature] Log metrics in test mode

  #1482 opened 4 months ago by mmeendez8
  8

• [Feature] remove AMP wrap in train_step

  #1457 opened 5 months ago by whlook
  5

• [Bug] DVCLiveVisBackend initializing DVC workspace in subdirectories

  #1508 opened 3 months ago by smarais
  1

• [Bug] Error Encountered with mmengine Dependency Involving JSON and Time Modules

  #1523 opened 2 months ago by Duguce
  0

• 不是大模型使用并行策略的效率大大降低 ！

  #1522 opened 2 months ago by Shen001
  1

• [Docs] Add OMG-Seg to ecoystem projects

  #1521 opened 2 months ago by evdcush
  0

• [Feature] Speed up the resume process of IterBased loop

  #1520 opened 2 months ago by YinAoXiong
  0

• [Bug] 性能问题：_get_valid_value函数首次调用torch.Tensor.item()时耗时过长

  #1519 opened 2 months ago by BenjaminPang
  0

• Suggested combination of Runner and AmpOptimWrapper does not result in mixed precision training [Docs]

  #1515 opened 2 months ago by JMQuehl
  0

• [Bug]

  #1514 opened 2 months ago by fsbarros98
  0

• [Feature] SegVisualizationHook: work in 'iter' but no 'epoch'

  #1513 opened 2 months ago by kanousen
  0

• Is anywhere record the version support matrix between mmengine and pytorch?

  #1509 opened 3 months ago by kelvinwang139
  2

• AttributeError: module 'torch.distributed' has no attribute 'ReduceOp'

  #1507 opened 3 months ago by AttilaLengyel-TomTom
  0

• [Bug] config to import yapf causes 'EOFError: Ran out of input' when distributed training

  #1480 opened 3 months ago by DeclK
  9

• [Bug] MMDistributedDataParallel distributed training can not help save memory. The total memory usage is twice that of a single card.

  #1504 opened 3 months ago by humian321
  2

• [Docs] Add "colossalai" in requirements

  #1487 opened 3 months ago by Yanjia0
  1

• scattering the data to gpu when using base dataelement

  #1501 opened 3 months ago by ajaynitk
  1

• [Feature] Nested initialization implementation of pure Python style configuration files

  #1467 opened 5 months ago by YinAoXiong
  3

• [Bug] load_from pretrained checkpoint fails using FlexibleRunner and DeepSpeed

  #1499 opened 3 months ago by pdmct
  2

• [Feature] Early Stopping, Validation Loss

  #1491 opened 3 months ago by 1dmesh
  1

• [Feature] A Checkpoint hook for saving model checkpoints as Weights & Biases Artifacts

  #1492 opened 3 months ago by soumik12345
  0

• [Feature] nn.LazyLinear

  #1484 opened 4 months ago by holdjun
  2

• [Docs] How do backend_args work?

  #1470 opened 4 months ago by Data-drone
  2

• activation_checkpointing 导致权重无法更新

  #1425 opened 5 months ago by Qidian213
  1

• 怎么调用ProfilerHook

  #1442 opened 5 months ago by ChaoyiXie
  1

• [Bug] Config to_dict() does not convert type recursively.

  #1464 opened 5 months ago by wangg12
  1

• KeyError: 'CocoDataset

  #1463 opened 5 months ago by lfreee
  1

• [Bug] TypeError: `logger` should be either a logging.Logger object, str, "silent", "current" or None, but got <class 'list'>

  #1452 opened 5 months ago by wang-tf
  3

• [Bug] MMDistributedDataParallel have no effect

  #1455 opened 5 months ago by doodoo0006
  4

• [Feature] 保存的pth文件越来越大

  #1446 opened 5 months ago by ChaoyiXie
  2

• [Feature] Log metrics to visualizer on test run

  #1450 opened 6 months ago by InakiRaba91
  0

• [Bug] Setting EpochBasedTrainLoop.val_begin=0 does not work!

  #1448 opened 6 months ago by ChanCody
  0

• [Feature] Serialize data list to torch.Tensor

  #1443 opened 6 months ago by wangg12
  0

• [Bug] error in readme

  #1439 opened 6 months ago by del-zhenwu
  0

• 'hybrid_parallel' plugin in 'ColossalAIStrategy' is not supported in mmengin-0.10.0

  #1432 opened 6 months ago by taohan10200
  1

• [Bug] NameError: name 'OptimWrapper' is not defined, when i used mmdeploy in jetson

  #1433 opened 6 months ago by lijoe123
  2

• [Bug] `scale_lr()` cannot be called after `ParamScheduler` in DDPStrategy using `FlexibleRunner`.

  #1427 opened 6 months ago by SCZwangxiao
  2

• [Bug] Deepcopy of BaseDataElement seems not working

  #1423 opened 7 months ago by RunsenXu
  0

• [Feature] Add xFormers to MMEngine

  #1420 opened 7 months ago by AkideLiu
  0

• [Bug] MMDeepSpeedEngineWrapper bf16 bug

  #1417 opened 7 months ago by felixfuu
  1