⬢github Python · 1.2K ★ +350 since we first saw it · pushed 8 h ago
lyramakesmusic/neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
A Python tool for mechanistic interpretability research on Qwen3.5-4B. It steers every MLP neuron in a layer up and down across prompts, records every rollout, and provides a local web explorer with a 3D map of neurons, steer/logit lenses, steered chat, UMAP embeddings, and a Claude-powered agent (via MCP) that can drive the model and interface live.
Why now: Trending on GitHub's most-starred new repos shortly after release, likely due to its unusual steering-first, keep-every-rollout approach and integrated AI-agent explorer.
Who it is for: Interpretability researchers and hobbyists who want to probe, steer, and visualize neuron-level behavior of a specific open model on their own hardware.
interpretabilitymechanistic-interpretabilityqwensteering
Stars over our 68 snapshots: 886 to 1.2K, since 17 h ago.
Where people talked about it
- ⬢github new repos, most starred 3 min ago
API: https://socialmediatrends-api.osmike.com/v1/repos/lyramakesmusic/neuron-explorer