Top postsTop post
Summary This is the third and final post in a series detailing our attempts to mechanistically interpret one head of the chess transformer Maia-3. We find that previous causal evidence gave us a spurious picture of the head, and that rather than just detecting knight forks, head 5 of layer...
Ablating one of Maia-3 23m chessformer's 128 attention heads destroys the policy toward the brilliant queen sacrifice. Uses https://github.com/chessformer-lens/chessformer_lens
Quick interp demo in colab: Localize knight forks to a single head in Maia-3 with logit-lens and per-head ablation. https://colab.research.google.com/drive/1YYZBd_SZbjOscRXIqJUbfCaY7rRbEzWx?usp=sharing (This is a demo of the library's capabilities so the sample size is tiny... much more analysis is done in an upcoming paper, for instance we mine hundreds of forks...
Summary of this post This is the second in a series of posts detailing my manifold findings while investigating how a chess transformer engine that mimics human play represents knight forks. Last post described strong correlational evidence that the knight-fork policy logit snaps into place after block 5’s attention layer....
I was poking around at a chessformer which mimics human play and made this fun companion app to visualize a lot of the internals of the engine. Check out how attention heads look, how the residual stream evolves, and play around with it if interested!
(parts 2 and 3 to follow) Summary of this post This post is on the results of a mechanistic interpretability project aimed at understanding the internals of Maia 3: a transformer based chess bot trained to imitate human play at a chosen skill level, rather than to play optimally. My...