[KongchangAI]
ConceptTemplate Lens

Template lens

A template-based activation-reading method that extracts the top-10 words or phrases at a specified layer as the readout result.

Timeline (last 90 days)

Oct 12

The activation-reading tools evaluated are divided into single-token readers and multi-token readers: the single-token side includes Logit lens, J-lens, R-lens, and Tuned lens; the multi-token side includes Natural Language Autoencoders, Sparse Autoencoders, Patchscopes, Template lens, and Oracle Lens.

Unverified50%

All Facts (1)

Source Articles