Comments (9)
你好,感谢你的关注与使用!
请问有具体的log信息吗?比如是在哪一个算子的处理过程当中产生的OOM错误,如能提供相关信息对我们定位具体原因有很大帮助~
此外,我们注意到你在处理时使用了较高的并行度(np=128),你可以将其减小后再进行尝试;以及注意到你开启了tracer,trace_num设置为了一个非常大的值,由于在tracer工作过程当中,有变化的样本是先存放在内存中然后导出到磁盘的,它需要占用额外的时间和内存,建议你这个值设置为一个较小的值,毕竟tracer只是帮助你去采样并了解OP的处理效果,
from data-juicer.
你好,感谢你的关注与使用!
请问有具体的log信息吗?比如是在哪一个算子的处理过程当中产生的OOM错误,如能提供相关信息对我们定位具体原因有很大帮助~
此外,我们注意到你在处理时使用了较高的并行度(np=128),你可以将其减小后再进行尝试;以及注意到你开启了tracer,trace_num设置为了一个非常大的值,由于在tracer工作过程当中,有变化的样本是先存放在内存中然后导出到磁盘的,它需要占用额外的时间和内存,建议你这个值设置为一个较小的值,毕竟tracer只是帮助你去采样并了解OP的处理效果,
感谢回复,具体log没有打印报错信息,观察内存占用很高导致Kill -9,,我做了几个实验尝试:
- 如您所说将并行度降低到了20,尝试后没有任何作用,还是OOM
- 之后将所有clean_links_mapper等相关mapper算子都去掉,只保留其余4个算子,可以正常运行,且内存占用在600G左右
from data-juicer.
将trace_num设置为一个较小的值(如默认的10)能否避免OOM呢?
from data-juicer.
将trace_num设置为一个较小的值(如默认的10)能否避免OOM呢?
设置了1000还是不行,内存基本没有太大变化
from data-juicer.
好的,请问你方便提供你的待处理的数据吗(可邮件告诉我们数据获取方式,邮箱: [email protected])?
不方便的话可以额外提供一下数据集的样本数目,以及这个数据集的大概类型(比如类似于common crawl之类的),以及别的一些你觉得有帮助的信息,然后我们这边尝试复现一下你遇到的这个问题
from data-juicer.
好的,请问你方便提供你的待处理的数据吗(可邮件告诉我们数据获取方式,邮箱: [email protected])?
不方便的话可以额外提供一下数据集的样本数目,以及这个数据集的大概类型(比如类似于common crawl之类的),以及别的一些你觉得有帮助的信息,然后我们这边尝试复现一下你遇到的这个问题
就是common crawl的一部分英文数据集,大概180G,总量在2500w条左右
from data-juicer.
嗨,我们这边对齐你这边的情况用CC的数据构造了一个数据集,大小189GB,包括25,359,505个数据样本
然后在一台754GB内存的机器上用与你的算子列表相同的流程对数据集进行处理,np设置为了48,tracer开启,trace_num设置为1000,结果能够正常处理完成。实验机器在未处理时内存水位为180GB左右(有其他实验在同时运行),基于此,你提到的4个mapper算子在处理的过程中峰值内存占用分别为:
- clean_email_mapper: 283GB
- clean_html_mapper: 288GB
- clean_ip_mapper: 282GB
- clean_links_mapper: 277GB
四个算子占用内存均为超过系统内存,均能够正常处理完成。因此我们这边未能复现出你遇到的OOM的问题。
清洗过程占用内存达1.4TB导致OOM
再请问你一下,你上文所说的这个“占用内存”具体指的是下面的哪一个?
- 随机访问存储RAM:在你的系统中可以使用
free -h
获取机器的总内存,上面的复现实验中的内存即指这个内存 - 硬盘存储:在你的系统中可以使用
df -h
获取机器中个磁盘/文件系统的存储使用情况
from data-juicer.
This issue is marked as stale because there has been no activity for 21 days. Remove stale label or add new comments or this issue will be closed in 3 day.
from data-juicer.
Close this stale issue.
from data-juicer.
Related Issues (20)
- DJ-v.0.2 docker image update HOT 1
- DJ-v0.2 API page enhancement
- Video content compliance and privacy protection operators (image, text, audio)
- [Bug]: video split by duration mapper return non-exist video
- support panda's student captioner model in our captioning mapper HOT 3
- [Bug]: Video_split_by_scene_mapper create non-exist video_keys
- [Feature Request] Implement more streamlined interfaces for users seeking minimal functionality (data_juicer.op.functional) HOT 2
- Request a sample code demonstrating the use of image_captioning_from_gpt4v_mapper.py HOT 3
- Can not download the data quality classifier models. HOT 1
- alphanumeric_filter算子清洗疑问 HOT 5
- Absolute path to relative path for multi-source
- [Bug]: process on ray occur "TypeError: 'str' object cannot be interpreted as an integer" HOT 8
- filter是否支持batch处理,以及怎么设置batch_size? HOT 5
- hash calculate in ray deduplicator HOT 4
- 为什么大部分的refined recipe都是用simhash去重? HOT 3
- [Bug]: 运行tools/analyze_data.py报错,出现 KeyError: 'text' HOT 2
- [Question] Can't find evalutor.yaml on the path of `/workspace/data-juicer/demos` HOT 1
- A Compatibility Issue in Environment Installation of DJ-Sandbox HOT 1
- stopwords_filter 为什么是过滤掉小于某个阈值的样本 HOT 3
- 报”error: Unrecognized arguments: -B -S -I -c“ HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from data-juicer.