What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

Neuronpedia Jacobian Lens: Interactive J-Space Explorer for Qwen3.6-27B

Connections

Related to

Generative Recursive Reasoning (GRAM) Interpretability tooling like the Jacobian Lens could be applied to probe the persistent latent states that GRAM's recursive reasoning models rely on.
Inkling: Thinking Machines' Open-Weights Model (and Tinker Fine-Tuning Platform) Jacobian-based interpretability methods developed for Qwen3.6-27B could plausibly extend to analyzing internal representations of large MoE models like Inkling.

Supported by

Qwen3.8 27B Model Analysis The Neuronpedia Jacobian Lens item specifically targets Qwen3.6-27B internals; 7ddd8e46's performance analysis contextualizes the capabilities of the same model architecture being mechanistically studied
Compression is prediction Mechanistic interpretability work like Jacobian Lens analysis implicitly relies on the compression-prediction equivalence — understanding how LLMs compress training data into weights explains what features like those in J-Space actually represent

Comparing 11 Different AI Models Both items investigate Qwen model internals and outputs — the 11-model comparison examines behavioral outputs while the Jacobian Lens explores Qwen3.6-27B's internal representations.
Qwen 3.8 2.4T Mixture of Experts Model Neuronpedia's Jacobian Lens tool is specifically designed to explore the internals of Qwen3 family models, making it a direct interpretability companion to this Qwen 3 MoE release
Understanding Embeddings in Language Models Neuronpedia's Jacobian Lens explores the internal representational space of transformer models — directly extending embedding fundamentals into mechanistic interpretability of how representations transform through layers
Exploring Claude/GPT Knowledge Cutoffs Both use interpretability/probing techniques to extract hidden internal information from large language models — Neuronpedia's Jacobian Lens explores internal representations, while the new item probes models to infer training metadata
Emergent Introspective Awareness in Large Language Models Both investigate internal model representations through activation-level analysis — Neuronpedia's Jacobian Lens explores feature spaces in Qwen3, while this paper injects concepts into activations to probe self-awareness, making them methodologically complementary mechanistic interpretability efforts.
Teaching Qwen to Paint with Code Both involve deep analysis of Qwen model internals — one probing Jacobian activations of Qwen3.6-27B, the other training Qwen's behavior via reinforcement learning

Type:	link
First published:	Monday, 6 July 2026 at 12:00 UTC
Location:	https://www.neuronpedia.org/qwen3.6-27b/jlens
URL:	https://mattwood.fyi/i/193e6d70

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections