Richard Jeevash

Mechanistic interpretability · Computational neuroscience

Richard Jeevash.

I like an explanation I can try to break. And a quiet scene that somehow breaks me.

I test whether explanations of neural networks hold up when you intervene on them, in language models and in models of the fly brain.

Findings go up before peer review, clearly labelled. Here’s why.

Fig. 00 / The basic moveRun AMaryRun BMary?copy one activation
Copy one activation from run A into run B. Does the answer move the way the explanation predicts?

The question running through it

Does the explanation survive the intervention?

  • Language models

    What an activation patch really moves, and when a test can tell a right explanation from a wrong one.

  • Models of the fly brain

    Wiring-based models of the fly’s brain: does silencing a cell type in the model match silencing it in the fly?

Latest findings

All research

Curiosity, with a paper trail.

Notes

All notes

Mostly coffee. Some questions.

  • Why I post findings before peer review

    Interpretability moves fast and much of it isn’t settled. Posting checked findings early, clearly labelled, seems more useful than holding them back for a year.

  • How I run a study

    Freeze the plan, save every output, generate every number from what was saved, and report what didn’t work. The routine behind each paper on this site.

Tools

  • Logogram

    A local-first workbench for causal experiments inside language models. Change one activation, watch what changes.

  • DigitalFly

    A virtual fruit fly you can poke: a simulated body steered by a brain you can inspect, silence and tune.