AssertionError: Default process group is not initialized

最新推荐文章于 2025-06-27 17:44:40 发布

日作诗歌三千首

最新推荐文章于 2025-06-27 17:44:40 发布

阅读量1.8k

点赞数 2

CC 4.0 BY-SA版权

本文链接：https://blog.youkuaiyun.com/u012483097/article/details/115911508

本文介绍了如何在PyTorch中设置基本的分布式训练环境。通过使用gloo后端并指定初始化方法、进程排名及世界大小等参数，可以实现单机多卡的简单分布式训练。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

在train文件中加入：
import torch.distributed as dist
dist.init_process_group('gloo', init_method='file:///tmp/somefile', rank=0, world_size=1)

确定要放弃本次机会？

福利倒计时

: :

立减 ¥

普通VIP年卡可用

立即使用

日作诗歌三千首

关注关注

2
点赞
踩
2

收藏

觉得还不错? 一键收藏
5
评论
分享

复制链接

分享到 QQ

分享到新浪微博

扫一扫
举报

举报

pytorch Default process group is not initialized

jacke121的专栏

06-27

2814

pytorch Default process group is not initialized

【Debug记录】AssertionError: Default process group is not initialized

Roaddd的博客

12-12

848

AssertionError: Default process group is not initialized

5 条评论您还未登录，请先登录后发表或查看评论

RuntimeError: Default process group has not been initialized, please make sure to call init_process_

最新发布

2303_79295792的博客

06-27

354

YOLOv7多卡训练报错 RuntimeError: Default process group has not been initialized，please make sure to call init_process_group.

pytorch-mmsegmentation train时遇到AssertionError:Default process group is not initialized

m0_37568067的博客

11-18

9186

问题原因：非分布式训练使用了分布式训练的设置两种解决办法： 1、在tools/train.py 中加入 import torch.distributed as dist dist.init_process_group('gloo', init_method='file:///temp/somefile', rank=0, world_size=1) 2、在configs/_base_/models 的文件中，首行norm_cfg = dict(type='SyncBN', requires_gr

pytorch分布式报错AssertionError: Default process group is not initialized

qq_40373651的博客

07-17

9674

pytorch分布式报错AssertionError: Default process group is not initialized 在pytorch中分布式中，dist.barrier()中报错AssertionError: Default process group is not initialized。可以尝试： import torch.distributed as dist dist.init_process_group('gloo', init_method='file:///tmp/so

assertionerror: default process group is not initialized

06-28

它通常意味着你尝试在一个未初始化的进程组中使用某个功能，因此Python引发了一个“AssertionError”异常。要解决这个问题，你需要先确保已经正确地初始化了进程组。你可以使用Python的multiprocessing模块来创建...

AssertionError: weight model.layers.0.self_attn.q_proj.weight does not exist

sqlora的专栏

11-08

1441

通义千问2.5-7B-Instruct-AWQ量化，但在npu上运行报上面错误，奇怪?： Exception:weight model.layers.0.self_attn.q_proj.weight does not exist AssertionError: weight model.layers.0.self_attn.q_proj.weight does not exist https://www.modelscope.cn/models/Qwen/Qwen2.5-7B-Instruc

Default process group has not been initialized, please make sure to call init_process_group.

qq_39483052的博客

10-16

2033

mmaction中的r2plus1d模型报错：Default process group has not been initialized, please make sure to call init_process_group. 在mmaction2-master/configs/base/models/r2plus1d_r34.py中，将norm_cfg=dict(type=‘SyncBN’, requires_grad=True, eps=1e-3)替换成norm_cfg=dict(type=‘BN3

Default process group has not been initialized, please make sure to call ini

weixin_41529093的博客

03-22

7230

仅作为记录，大佬请跳过。将distributed的值改为False，即可。参考传送门

Default process group has not been initialized, please make sure to call init_process_group可能解决方法

york1996的博客

11-05

3851

syncBN改成BN

RuntimeError: Default process group has not been initialized,please make sure to call init_process_g

qq_46024883的博客

06-08

5561

一直报错RuntimeError: Default process group has not been initialized,please make sure to call init_process_group.