About alignment - Githubissues

thuiar / MMSA

MMSA is a unified framework for Multimodal Sentiment Analysis.

MIT License

642 stars 104 forks source link

Closed JJYdesu closed 2 years ago

JJYdesu commented 2 years ago

hello! I want to know that for MOSI and MOSEI, how you align three modalities? thank you !

FlameSky-S commented 2 years ago

We don't do alignments in this framework. The aligned features are provided by the CMU team. Please refer to their papers for more information.