ERNIE 5.0: A 2.4 Trillion-Parameter Unified Multimodal Foundation Model

• multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.

ERNIE 5.0 is a 2.4-trillion-parameter unified multimodal model trained from scratch across text, image, video, and audio, avoiding late-fusion architecture limitations.

• articleAnnouncing the Open Source Release of the ERNIE 4.5 Model Familyunknown
• articleIntroducing ERNIE-Imageunknown
• articleERNIE-4.5-VL-28B-A3B-Thinking: A Breakthrough in Multimodal AIunknown

Checking sign-in…

Loading comments…