-
Notifications
You must be signed in to change notification settings - Fork 667
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OneFlow 算子对齐 PyTorch 完备计划推进表 #4936
Comments
This comment was marked as outdated.
This comment was marked as outdated.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment was marked as outdated.
This comment was marked as outdated.
Merged
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This was
linked to
pull requests
Jun 7, 2021
Merged
Merged
This comment has been minimized.
This comment has been minimized.
This comment was marked as outdated.
This comment was marked as outdated.
Closed
This comment was marked as outdated.
This comment was marked as outdated.
Closed
Closed
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
为什么einsum 这个没有在列表里 |
This comment was marked as outdated.
This comment was marked as outdated.
2 tasks
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
背景
本 issue 主贴(此贴)记录分工和对应的 PR,方便跟进。
解决的时间节点
Hard难度的应该在0.7.0版本之前解决,Medium应该在0.6.0版本之前完全解决,Easy难度的应该在分配当周或者下周进行解决。
相关文档
负责人
@BBuf
任务分工表
flow.sum()仅支持dim指定维度,而torch支持dim或axis。来源:迟子秋。https://github.com/Oneflow-Inc/OneTeam/issues/1184#issuecomment-1089842163oneflow._oneflow_internal.Tensor 类型不可以自动转化为Numpy类型 来源:陈巧玲 OCR https://github.com/Oneflow-Inc/OneCloud/issues/69#issuecomment-1074628723Dataloader pin_memory没有对齐的问题。需要露阳解决。优先级低。相关issue:https://github.com/Oneflow-Inc/OneTeam/issues/1180 。将argsort的Python层逻辑迁移到Functor。来源性能需求。认领人:王世杰。 Medium。将argwhere的Python层逻辑迁移到Functor。来源性能需求。认领人:王世杰。 Medium。将to_local_op的Python层逻辑迁移到Functor。来源性能需求。Easy。将empty的Python层逻辑迁移到Functor。来源性能需求。 Medium。flow.cuda.empty_cache 接口没对齐。来源:林松。https://github.com/Oneflow-Inc/OneCloud/issues/87#issuecomment-1102115749 。来自丰伟:allocator需要通过vm从stream中获取。当前支持的条件不成熟,且功能较为薄弱。我们是从该axis到最后,做norm。对应的解决方案可以是前后加2个transpose。#787 。认领人:刘沛宏。 Medium。
tensor.median()
方法,来源:任天和。期望libai mae支持graph格式数据并行,流水线并行和模型并行 libai#259 。认领人:王世杰。dev_median #8069torch.cuda.synchronize
以及torch.cuda.max_memory_alocated
接口。来源: 任天和。期望libai mae支持graph格式数据并行,流水线并行和模型并行 libai#259torch.nn.parallel.DistributedDataParallel()
的入参没有对齐。来源:任天和。期望libai mae支持graph格式数据并行,流水线并行和模型并行 libai#259oneflow.nn.utils.clip_grad_norm_
不支持传入None,来源:任天和。期望libai mae支持graph格式数据并行,流水线并行和模型并行 libai#259#6156
The text was updated successfully, but these errors were encountered: