如何选择FairSeq选项`-ddp-backend`

发布于 2025-02-09 02:57:57 字数 880 浏览 2 评论 0 原文

我正在学习如何使用 fairseq 实现基于简单的翻译模型在变压器上。

我想在实验室服务器上使用2 GEFORCE RTX 3090 GPU。我应该选择的哪个选项 - fairseq-train 的ddp-backend

此外,您能否解释的所有以下选项的含义-DDP-Backend 以及何时使用它们的含义?

来自 FairSeq-Train => distribute_training


- DDP-BAKEND

可能的选择:c10d,firls_sharded,legacy_ddp,no_c10d,pytorch_ddp,slowmo

distribationdataparallear后端

默认值:“ pytorch_ddp”

distributeddataPararallearpallelallearlallearallearlallelalearder 有任何不当行动。

I'm learning how to use fairseq to implement a simple translation model based on Transformer.

I would like to use 2 GeForce RTX 3090 GPUs on my lab server. Which option should I select for --ddp-backend of fairseq-train?

Furthermore, could you explain about the meaning of all following options for --ddp-backend and when to use them respectively?

From fairseq Documentation: Command-line Tools => fairseq-train => distributed_training


--ddp-backend:

Possible choices: c10d, fully_sharded, legacy_ddp, no_c10d, pytorch_ddp, slowmo

DistributedDataParallel backend

Default: “pytorch_ddp”

I'm new to stack exchange community, sorry if there is any inappropriate action.

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(2

嘦怹 2025-02-16 02:57:57

我不太确定,但是我在

DDP_BACKEND_CHOICES = ChoiceEnum(
    [
        "c10d",  # alias for pytorch_ddp
        "fully_sharded",  # FullyShardedDataParallel from fairscale
        "legacy_ddp",
        "no_c10d",  # alias for legacy_ddp
        "pytorch_ddp",
        "slowmo",
    ]
)

可能会有所帮助,但我也在努力

I am not too sure, but I found this on GitHub

DDP_BACKEND_CHOICES = ChoiceEnum(
    [
        "c10d",  # alias for pytorch_ddp
        "fully_sharded",  # FullyShardedDataParallel from fairscale
        "legacy_ddp",
        "no_c10d",  # alias for legacy_ddp
        "pytorch_ddp",
        "slowmo",
    ]
)

Might be helpful, but I am also struggling with this

被你宠の有点坏 2025-02-16 02:57:57

您可以在options.py文件中找到它,希望它有用。但是他们只描述了“ C10D”和“ NO_C10D”之间的区别。因此,我们应该继续寻找更多。

这是

You can find this in the options.py file, hope it's helpful. But they only describe the difference between "c10d" and "no_c10d". So we should keep going to find more.

This is the link.

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文