Went through it but zero mention of an obvious contender: data shape. Latest models like Qwen-3 or Kimi are trained on vast amounts of "clean" synthetic data/environments and this impacts superposition/features.
Data shape may explain superposition in Qwen-3 and Kimi models
AISummary
Alexander Doria says interpretability analyses of modern models overlook data shape as a likely cause of superposition. He argues that Qwen-3 and Kimi, trained on large amounts of clean synthetic data and environments, may have features and superposition shaped by that training data.
Post on XView on X
Alexander DoriaVerified on X
@Dorialexander
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works https://already-clean.goodfire.com/View quoted post on X
Source: Alexander Doria · x.comPublished · added here