{"type":"post","id":"neurons-as-an-interpretable-basis-ffb4bb","slug":"neurons-as-an-interpretable-basis-ffb4bb","title":"Neurons as an Interpretable Basis","description":"Goodfire reports experiments on Qwen3.6-35B-A3B suggesting that individual MLP neurons can be treated as interpretable computational units.","retrieval_nugget":"Goodfire reports experiments on Qwen3.6-35B-A3B suggesting that individual MLP neurons can be treated as interpretable computational units. Many weak activations could be removed with modest effects, while concise descriptions of strong activity supported prediction and simulation. For builders, the contribution is a testable method for turning explanations of model internals into causal hypotheses.","published_at":"2026-11-11","updated_at":"2026-11-11","record_date":"2026-11-11","date_kind":"published_at","topics":["models"],"entities":["Goodfire","Qwen3.6-35B-A3B","neuron-explorer"],"related_patterns":[],"novelty_label":"new_fact","what_changed":"Goodfire reports experiments on Qwen3.6-35B-A3B suggesting neurons are interpretable computational units, with sparse activation, few legible modes, and agent-generated descriptions that predict neuron behavior.","prior_refs":[],"source_url":"https://www.goodfire.com/research/neurons-as-an-interpretable-basis","source_urls":["https://www.goodfire.com/research/neurons-as-an-interpretable-basis"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:neurons-as-an-interpretable-basis-ffb4bb","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/neurons-as-an-interpretable-basis-ffb4bb/","markdown":"https://newruntime.com/posts/neurons-as-an-interpretable-basis-ffb4bb.md","json":"https://newruntime.com/posts/neurons-as-an-interpretable-basis-ffb4bb.json"},"markdown_url":"https://newruntime.com/posts/neurons-as-an-interpretable-basis-ffb4bb.md","next_reads":[{"type":"topic","path":"/topics/models/","url":"https://newruntime.com/topics/models/","title":"Models","reason":"Explore the models topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/models.md","json_url":"https://newruntime.com/topics/models.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}