Multimodal learning - Wikipedia
5
公开标注数
5
参与人数
2026-07-25 09:40:14
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-25 13:01:14
原文高亮摘录
“Models such as CLIP (Contrastive Language–Image Pretraining) learn joint representations of images and text by optimizing contrastive objectives, allowing the model to match images with…”
Whisper 随想笔记
So basically it's like a game of matching pairs, makes sense now.
划选高亮2026-07-25 12:52:14
原文高亮摘录
“Models such as CLIP (Contrastive Language–Image Pretraining) learn joint representations of images and text by optimizing contrastive objectives, allowing the model to match images with…”
Whisper 随想笔记
CLIP's pretty cool but contrastive learning can be a bit tricky to wrap your head around.
划选高亮2026-07-25 09:58:14
原文高亮摘录
“Multimodal learning is a type of deep learning that integrates and processes multiple types of data”
Whisper 随想笔记
I've seen this in action for image captioning, it's pretty cool actually.
划选高亮2026-07-25 09:49:14
原文高亮摘录
“Multimodal learning is a type of deep learning that integrates and processes multiple types of data”
Whisper 随想笔记
Doesn't this just complicate things? Sometimes one modality is enough.
划选高亮2026-07-25 09:40:14
原文高亮摘录
“Multimodal learning is a type of deep learning that integrates and processes multiple types of data”
Whisper 随想笔记
So it's basically deep learning with extra senses, makes sense.
分享本页 Whisper
短链接
https://domwhisper.com/s/85a07e02efe9嵌入代码
<iframe src="https://domwhisper.com/embed/85a07e02efe9" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>