sleepychord / cogdata Goto Github PK
View Code? Open in Web Editor NEWA light-weight data management system for large-scale pretraining
Home Page: https://sleepychord.github.io/cogdata/
License: MIT License
A light-weight data management system for large-scale pretraining
Home Page: https://sleepychord.github.io/cogdata/
License: MIT License
when I run cogdata process --task_id test_task --nproc 2 --dataloader_num_workers 1 --ratio 1 test_ds
in the terminal, I got a different result from the guide.
And when I try to view the log via tail -f /mnt/disk/reself/projects/CogView2/cogdata_task_test_task/main_pid_46834.log
, it appears as follow:
how can I fix it?
我按照cogview的README,使用vqvae_hard_biggerset_011.pt
处理一个1300多张的小型图片数据集,分辨率为256*256
,但我使用单卡P100显示
RuntimeError: CUDA out of memory. Tried to allocate 16.00 GiB (GPU 0; 15.90 GiB total capacity; 856.64 MiB already allocated; 14.07 GiB free; 874.00 MiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
所以我比较好奇处理这个小数据集需要多大的显存
[background]
I want to use my own text-image datasets to generate binary format dataset for CogView training in 'https://github.com/THUDM/CogView'. It has been mentioned in that repo the author use this cogdate toolkit to preprocess data.
[question]
Would you please tell me how to organize my raw text-image dataset, and then how to use the cogdata toolkit to generate the target bin file? for example, whether i should name the a text-image pair the same, such as 'a dog sits on the ground.txt' and 'a dog sits on the ground.png', or i should take other forms?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.