Alexandre Variengien

My guess at Con­jec­ture’s vi­sion: trig­ger­ing a nar­ra­tive bifurcation

The case for train­ing fron­tier AIs on Sume­rian-only corpus

A Univer­sal Emer­gent De­com­po­si­tion of Retrieval Tasks in Lan­guage Models

Cap­ture the Flag Mechanis­tic In­ter­pretabil­ity Challenges

In­put Swap Graphs: Dis­cov­er­ing the role of neu­ral net­work com­po­nents at scale

An in­tro­duc­tion to lan­guage model interpretability

Some com­mon con­fu­sion about in­duc­tion heads

Gliders in Lan­guage Models

Some Les­sons Learned from Study­ing Indi­rect Ob­ject Iden­ti­fi­ca­tion in GPT-2 small

Ap­ply to the Ma­chine Learn­ing For Good boot­camp in France

Croe­sus, Cer­berus, and the mag­pies: a gen­tle in­tro­duc­tion to Elic­it­ing La­tent Knowledge