open-mmlab/mmengine issues and pull requests

#1584 - [Feature] Request for Multi-Input Support in BaseModel

Issue - State: open - Opened by shenshanf 21 days ago

#1583 - [Feature] Support for XPU devices

Issue - State: open - Opened by iori2333 22 days ago - 1 comment

#1582 - [Feature] add_scalar in logger_hook for after_test_epoch

Issue - State: open - Opened by AwePhD about 1 month ago

#1581 - [Bug] “please pass a valid scope. ” in the thirdpart project

Issue - State: open - Opened by zhanzf about 1 month ago
Labels: bug

#1579 - [Bug] Fix a MLflow vis backend bug when the config key is not `str`

Pull Request - State: open - Opened by zgjja about 1 month ago - 1 comment

#1578 - Update metric.py，support multi-server training with Ascend 910B

Pull Request - State: open - Opened by barondai about 1 month ago - 1 comment

#1577 - Allow Torch half models

Pull Request - State: open - Opened by ProgramadorArtificial about 1 month ago - 1 comment

#1576 - Defaults to utf-8 encoding

Pull Request - State: open - Opened by huaibovip about 1 month ago

#1575 - [Bug] test suite fails with `...AttributeError: '<SOME_CLASS>' object has no attribute 'runTest'`

Issue - State: open - Opened by GaetanLepage about 2 months ago
Labels: bug

#1574 - [Bug] MLflowVisBackend crashed on after_run hook

Issue - State: open - Opened by maxvandenhoven about 2 months ago - 1 comment
Labels: bug

#1573 - [Feature] Support for torch > 2.1

Issue - State: open - Opened by wemoveon2 about 2 months ago

#1572 - Yehc/bump version to v0.10.5

Pull Request - State: closed - Opened by HAOCHENYE about 2 months ago - 1 comment

#1571 - [Fix] Fix check builtin module

Pull Request - State: closed - Opened by HAOCHENYE about 2 months ago

#1570 - Supports mps device IndexType

Pull Request - State: open - Opened by bibekyess 2 months ago - 1 comment

#1568 - [Docs] Is there way to extract model from mmengine back to Pytorch

Issue - State: open - Opened by MichaelD7911 2 months ago

#1567 - 为了方便技术交流，拉了一个多模态大模型技术交流群，有需要的大家可以加入

Issue - State: closed - Opened by feihuamantian 2 months ago

#1566 - [Feature] Modifed implementation of BaseDataElement cuda(), cpu(), etc

Issue - State: open - Opened by YangyangFu 3 months ago

#1565 - [Docs] after resume training, change param_scheduler is no effective

Issue - State: open - Opened by zhangchao-s 3 months ago - 1 comment

#1564 - Several Improvements for the latest PyTorch Framework

Pull Request - State: open - Opened by MGAMZ 4 months ago - 2 comments

#1563 - [Feature] Add a parameter for selecting image suffixes in the LocalVisBackend class

Issue - State: open - Opened by chenller 4 months ago

#1562 - [Docs] A100算力加持！书生大模型实战营第3期全面升级，趣味闯关模式等你开启

Issue - State: open - Opened by boshallen 4 months ago

#1560 - What parameters are supported by save best

Issue - State: open - Opened by luoyq6 4 months ago - 2 comments

#1560 - What parameters are supported by save best

Issue - State: open - Opened by luoyq6 4 months ago - 2 comments

#1559 - [Fix] add Config hyperparameters for clearml backend

Pull Request - State: open - Opened by gntoni 4 months ago - 1 comment

#1559 - [Fix] add Config hyperparameters for clearml backend

Pull Request - State: open - Opened by gntoni 4 months ago - 1 comment

#1559 - [Fix] add Config hyperparameters for clearml backend

Pull Request - State: open - Opened by gntoni 4 months ago - 1 comment

#1558 - fix import torchvision when check_env

Pull Request - State: closed - Opened by zhyncs 4 months ago - 2 comments

#1558 - fix import torchvision when check_env

Pull Request - State: closed - Opened by zhyncs 4 months ago - 2 comments

#1557 - [Bug] `misaligned address` during in `SyncBuffersHook` all_reduce when using bf16 with deepspeed

Issue - State: open - Opened by SCZwangxiao 5 months ago
Labels: bug

#1556 - [Bug] KeyError: 'SentryFishVideo3Dataset is not in the main::datasets registry. Please check whether the value of `SentryFishVideo3Dataset` is correct or it was registered as expected. More details can be found at https://mmengine.readthedocs.io/en/latest/advanced_tutorials/config.html#import-the-custom-module'

Issue - State: open - Opened by seamoonlight-YBY 5 months ago
Labels: bug

#1555 - [Feature] 支持VisualDL可视化分析工具

Issue - State: open - Opened by chenller 5 months ago

#1553 - [Bugs] Disable grad scaler in AmpOptimWrapper when dtype is bf16

Pull Request - State: open - Opened by HIT-cwh 5 months ago

#1550 - Add progress bar for advancing dataloader when resume training

Pull Request - State: open - Opened by huuquan1994 5 months ago - 1 comment

#1549 - [Docs] 分布式训练和GPU的选择能不能在配置文件中读取？

Issue - State: open - Opened by anzitong 5 months ago

#1548 - [Fix] Speed up "--resume"

Pull Request - State: open - Opened by chtzs 6 months ago - 4 comments

#1546 - [Feature] 怎样得到每个类别的精确率和召回率等指标？

Issue - State: closed - Opened by anzitong 6 months ago - 1 comment

#1545 - [Feature] 训练得到的权重命名问题

Issue - State: closed - Opened by anzitong 6 months ago - 2 comments

#1543 - [Feature] about iterabledataset or webdataset

Issue - State: open - Opened by KyanChen 6 months ago - 1 comment

#1542 - [Bug] lr is wrong when resume with an LinearLR

Issue - State: closed - Opened by Cosmo1210 6 months ago - 1 comment
Labels: bug

#1541 - FSDPStrategy how to set mixed_precision and other params of pytorch

Issue - State: open - Opened by apachemycat 6 months ago

#1540 - [Bug] 分布式训练代码例子报错，

Issue - State: closed - Opened by apachemycat 6 months ago - 5 comments
Labels: bug

#1539 - ValueError: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. T

Issue - State: open - Opened by apachemycat 6 months ago - 1 comment
Labels: bug

#1538 - [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!

Issue - State: open - Opened by Helen-Ren-yi 6 months ago
Labels: bug

#1537 - Update is_mlu_available

Pull Request - State: closed - Opened by ClowDragon 6 months ago - 4 comments

#1536 - [Bug] 多卡情况下，训练后eval和离线test的精度不能保证一致

Issue - State: open - Opened by whlook 6 months ago
Labels: bug

#1535 - [Feature] Support dataset streaming?

Issue - State: open - Opened by Ablustrund 6 months ago

#1534 - Bump version to 0.10.4

Pull Request - State: closed - Opened by fanqiNO1 7 months ago

#1533 - [Bug] Unable to save results using pklfile_prefix tag

Issue - State: open - Opened by abadithela 7 months ago - 1 comment
Labels: bug

#1532 - fix typo

Pull Request - State: closed - Opened by hiramf 7 months ago - 1 comment

#1531 - [fix] Handle case ClearML already been initialized

Pull Request - State: closed - Opened by mhtrinh 7 months ago - 2 comments

#1529 - [Improve] Perform evaluation upon training completion

Pull Request - State: closed - Opened by LZHgrla 7 months ago

#1527 - [Fix] SLURM distributed training in containers where `scontrol` is not available

Pull Request - State: open - Opened by R-Fehler 7 months ago - 1 comment

#1525 - [Bug] No module named 'mmengine.models'

Issue - State: open - Opened by hitbuyi 7 months ago - 2 comments
Labels: bug

#1523 - [Bug] Error Encountered with mmengine Dependency Involving JSON and Time Modules

Issue - State: open - Opened by Duguce 7 months ago - 3 comments
Labels: bug

#1522 - 不是大模型使用并行策略的效率大大降低！

Issue - State: closed - Opened by Shen001 7 months ago - 2 comments

#1521 - [Docs] Add OMG-Seg to ecoystem projects

Issue - State: closed - Opened by evdcush 8 months ago

#1520 - [Feature] Speed up the resume process of IterBased loop

Issue - State: open - Opened by YinAoXiong 8 months ago - 2 comments

#1519 - [Bug] 性能问题：_get_valid_value函数首次调用torch.Tensor.item()时耗时过长

Issue - State: open - Opened by BenjaminPang 8 months ago
Labels: bug

#1518 - DeepSpeed2 不能自动排除冻结的参数[Bug]

Issue - State: closed - Opened by Baboom-l 8 months ago - 1 comment
Labels: bug

#1517 - [Enhance] Enable `exclude_frozen_parameters` for `DeepSpeedEngine._zero3_consolidated_16bit_state_dict`

Pull Request - State: closed - Opened by LZHgrla 8 months ago

#1515 - Suggested combination of Runner and AmpOptimWrapper does not result in mixed precision training [Docs]

Issue - State: open - Opened by JMQuehl 8 months ago

#1514 - [Bug]

Issue - State: open - Opened by fsbarros98 8 months ago
Labels: bug

#1513 - [Feature] SegVisualizationHook: work in 'iter' but no 'epoch'

Issue - State: open - Opened by kanousen 8 months ago

#1512 - Add the supported pytorch versions in README

Pull Request - State: closed - Opened by zhouzaida 8 months ago

#1511 - Fix config of readthedocs

Pull Request - State: closed - Opened by zhouzaida 8 months ago - 1 comment

#1509 - Is anywhere record the version support matrix between mmengine and pytorch?

Issue - State: closed - Opened by kelvinwang139 8 months ago - 2 comments

#1508 - [Bug] DVCLiveVisBackend initializing DVC workspace in subdirectories

Issue - State: open - Opened by smarais 8 months ago - 1 comment
Labels: bug

#1507 - AttributeError: module 'torch.distributed' has no attribute 'ReduceOp'

Issue - State: open - Opened by AttilaLengyel-TomTom 8 months ago
Labels: bug

#1506 - [Doc]: Fix docstring of Config

Pull Request - State: closed - Opened by MambaWong 9 months ago - 1 comment

#1505 - [Feature] Support custom `artifact_location` in MLflowVisBackend.

Pull Request - State: closed - Opened by daavoo 9 months ago - 2 comments

#1504 - [Bug] MMDistributedDataParallel distributed training can not help save memory. The total memory usage is twice that of a single card.

Issue - State: closed - Opened by humian321 9 months ago - 2 comments
Labels: bug

#1503 - [Feature] Support calculating loss during validation

Pull Request - State: closed - Opened by fanqiNO1 9 months ago - 1 comment

#1502 - Update `WandbVisBackend`

Pull Request - State: open - Opened by soumik12345 9 months ago

#1501 - scattering the data to gpu when using base dataelement

Issue - State: closed - Opened by ajaynitk 9 months ago - 1 comment

#1499 - [Bug] load_from pretrained checkpoint fails using FlexibleRunner and DeepSpeed

Issue - State: closed - Opened by pdmct 9 months ago - 2 comments
Labels: bug

#1498 - [Feature] A Checkpoint hook for saving model checkpoints as Weights & Biases Artifacts

Pull Request - State: open - Opened by soumik12345 9 months ago - 1 comment

#1497 - Fix fileio backend ut

Pull Request - State: closed - Opened by zhouzaida 9 months ago

#1496 - Remove codeowners file

Pull Request - State: closed - Opened by zhouzaida 9 months ago

#1495 - Move checkpoint funtions from runner to a new sub-package

Pull Request - State: open - Opened by zhouzaida 9 months ago

#1494 - [Fix] Fix warning capture

Pull Request - State: closed - Opened by fanqiNO1 9 months ago

#1492 - [Feature] A Checkpoint hook for saving model checkpoints as Weights & Biases Artifacts

Issue - State: open - Opened by soumik12345 9 months ago

#1491 - [Feature] Early Stopping, Validation Loss

Issue - State: open - Opened by 1dmesh 9 months ago - 1 comment

#1490 - User can choose device in cfg. If not, fall back to current protocol.

Pull Request - State: open - Opened by VincentCoulombe 9 months ago - 2 comments

#1488 - Fix typos and remove fullwidth unicode chars

Pull Request - State: closed - Opened by evdcush 9 months ago - 4 comments

#1487 - [Docs] Add "colossalai" in requirements

Issue - State: closed - Opened by Yanjia0 9 months ago - 1 comment

#1486 - [Feature] Support calculating loss in the validation step

Issue - State: closed - Opened by zhouzaida 9 months ago - 2 comments

#1484 - [Feature] nn.LazyLinear

Issue - State: open - Opened by holdjun 9 months ago - 2 comments
Labels: bug

#1482 - [Feature] Log metrics in test mode

Issue - State: open - Opened by mmeendez8 9 months ago - 8 comments

#1481 - fix typo

Pull Request - State: closed - Opened by zhouzaida 10 months ago

#1480 - [Bug] config to import yapf causes 'EOFError: Ran out of input' when distributed training

Issue - State: closed - Opened by DeclK 10 months ago - 9 comments
Labels: bug

#1479 - Refine mmengine intro

Pull Request - State: closed - Opened by zhouzaida 10 months ago

#1478 - Bump version to 0.10.3

Pull Request - State: closed - Opened by zhouzaida 10 months ago

#1476 - Update visualizer.py

Pull Request - State: closed - Opened by Anm-pinellia 10 months ago - 1 comment

#1474 - [Feature] Support `save_optimizer=False` for DeepSpeed

Pull Request - State: closed - Opened by LZHgrla 10 months ago

#1473 - remove deepcopy call to allow ONNX export

Pull Request - State: open - Opened by prathikr 10 months ago - 1 comment

#1471 - [Fix] Fix the resume of iteration

Pull Request - State: closed - Opened by LZHgrla 10 months ago

#1470 - [Docs] How do backend_args work?

Issue - State: open - Opened by Data-drone 10 months ago - 2 comments

#1469 - [Fix] Fix `dist.collect_results` to keep all ranks' elements

Pull Request - State: closed - Opened by LZHgrla 10 months ago

#1467 - [Feature] Nested initialization implementation of pure Python style configuration files

Issue - State: open - Opened by YinAoXiong 10 months ago - 3 comments

#1466 - [Docs] Add the usage of ProfilerHook

Pull Request - State: closed - Opened by zhouzaida 10 months ago

GitHub / open-mmlab/mmengine issues and pull requests