[MMC] implement multiple dataset with round robin in pretraining

facebookresearch / mmf

A modular framework for vision & language multimodal research from Facebook AI Research (FAIR)

https://mmf.sh/

Other

5.46k stars 932 forks source link

[MMC] implement multiple dataset with round robin in pretraining #1192

Closed BruceChaun closed 2 years ago

BruceChaun commented 2 years ago

Summary: Merge two pretraining datasets and support round robin strategy. Each loss is computed given a certain dataset. This slightly changes the MMFLoss where it skips the computation if the current dataset name is not in the specified datasets.

Reviewed By: lichengunc

Differential Revision: D33834556

facebook-github-bot commented 2 years ago

This pull request was exported from Phabricator. Differential Revision: D33834556

facebook-github-bot commented 2 years ago

This pull request was exported from Phabricator. Differential Revision: D33834556