I Built a Multimodal Embedding Model From Scratch on an RTX 4060 (Text, Image, Audio, and Video…
中文摘要
作者分享了如何仅凭单块 RTX 4060 显卡复现 Jina AI 的 GELATO 多模态嵌入模型架构，并记录了多次实验过程。
English Summary
The author describes reproducing Jina AI's GELATO multimodal embedding architecture on a single RTX 4060 GPU, detailing the experimental process and failures.
Original Excerpt
How I reproduced the architecture behind Jina AI’s GELATO on a single consumer laptop GPU, and what seven rounds of failed experiments… Continue reading on Medium »