search

OFA-Sys / ONE-PEACE

A general representation model across vision, audio, language modalities. Paper: ONE-PEACE: Exploring One General Representation Model Toward Unlimited Modalities
Apache License 2.0
964 stars 63 forks source link

GPU Device Config #20

Closed 3xWangDot closed 1 year ago

3xWangDot commented 1 year ago

Thanks for sharing this project and codebase. May I know the GPU device config you used in the pre-trained Stage1/2 ? (e.g. what type of GPU and how many of them) I didn't find it in the paper and project page. Thanks.

logicwong commented 1 year ago

@3xWangDot We trained ONE-PEACE on 256 A100 GPU with 80GB of RAM (256 for Stage1 and 32 for Stage2). The training process took approximately 20 days.