This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
14 Likes
Reward
14
6
Repost
Share
Comment
0/400
AirdropCollector
· 8h ago
The dimensionality reduction attack is extremely powerful.
FastVLM is here to show you something different! ✨
They attached MLP to FastViTHD and derived visual tokens in the LLM world.
Result? The number of tokens has significantly decreased, 4 times less than FastViT, 16 times less than ViT‑L/14, resolution 336 pixels!😲
The tokens have decreased, and the complexity has also lowered; it's simply the successful weight loss of the token world! 🤣📉