Website Text

<aside> 💡 30 words max.

</aside>

Researchers will investigate ways to compress and distill pretrained models (like BERT, GPT-3 and ViLBERT) without compromising their accuracy.