LESSWRONG
LW

Shan23Chen
14Ω2320
Message
Dialogue
Subscribe

shanchen.dev

Posts

Sorted by New
8Are SAE features from the Base Model still meaningful to LLaVA?
Ω
3mo
Ω
2
5Sparse Autoencoder Features for Classifications and Transferability
3mo
0
5Are SAE features from the Base Model still meaningful to LLaVA?
5mo
2

Wikitag Contributions

No wikitag contributions to display.

Comments

Sorted by
Newest
Are SAE features from the Base Model still meaningful to LLaVA?
Shan23Chen3mo10

thank you! full paper and code will be ready soon!

Reply
Are SAE features from the Base Model still meaningful to LLaVA?
Shan23Chen4mo10

Thank you! And thanks for making llava-gemma!

Reply