Commit History

Update autointerp results (first 900 latents, tighter CIs)
e251d68
verified

luciaquirke commited on

Update autointerp results (first 800 latents, tighter CIs)
a2eff67
verified

luciaquirke commited on

Update autointerp results (first 700 latents, tighter CIs)
a50c921
verified

luciaquirke commited on

Update autointerp results (first 600 latents, tighter CIs)
36ddaca
verified

luciaquirke commited on

Update autointerp results (first 500 latents, tighter CIs)
5f3bd3d
verified

luciaquirke commited on

Update autointerp results (first 400 latents, tighter CIs)
e017bf0
verified

luciaquirke commited on

Update autointerp results (first 300 latents, tighter CIs)
1e32eb8
verified

luciaquirke commited on

Update autointerp results (first 200 latents, tighter CIs)
25afd41
verified

luciaquirke commited on

Add delphi automated-interpretability results (200M-token protocol, detection+fuzzing with bootstrap CIs)
7e1ff03
verified

luciaquirke commited on

Trim README
cbbbc38
verified

luciaquirke commited on

Upload SmolLM2-1.7B layer17 SAE (adam, lr=1e-4)
63ef4fa
verified

luciaquirke commited on

initial commit
532f001
verified

luciaquirke commited on