neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
- 2k
- Python
- MIT
- 3 wk ago
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
Turn a knob inside a small open model instead of writing a prompt. A reproducible RepE and CAA steering harness with an honest benchmark, including the failures.
FEGA, a pipeline for analyzing how sparse autoencoder features causally affect language-model behavior and representation geometry.
This repository hosts Gardener Enhancement Proposals (GEPs): design and proposal documents for project-wide changes in Gardener that require discussion and decisions in the Product Steering Committee (PSC) and/or Technical Steering Committee (TSC).
Dose-calibrated block-sparse-dictionary steering for Qwen3.5-2B — interactive demo + preregistered head-to-head vs SAE features