---
title: Mechanistic Interpretability
description: AI notes by Mrinaal Arora tagged "Mechanistic Interpretability"
updated: 2026-07-30
canonical: https://aroramrinaal.com/ai/tags/mechanistic-interpretability
source: https://aroramrinaal.com/ai/tags/mechanistic-interpretability.md
alternate: https://aroramrinaal.com/ai/tags/mechanistic-interpretability/markdown
---

# Mechanistic Interpretability

AI notes, small builds, and writeups by Mrinaal Arora tagged "Mechanistic Interpretability".

- Page: https://aroramrinaal.com/ai/tags/mechanistic-interpretability
- Markdown: https://aroramrinaal.com/ai/tags/mechanistic-interpretability/markdown
- Markdown (.md): https://aroramrinaal.com/ai/tags/mechanistic-interpretability.md
- All tags: https://aroramrinaal.com/ai/tags/markdown
- Full AI index: https://aroramrinaal.com/ai/markdown
- Posts with this tag: 2 of 15
- Co-occurring tags: Sparse Autoencoders, Trustworthy AI

## Posts

| Post | Date | Tags | Word count | Source |
| --- | --- | --- | --- | --- |
| [Before It Codes: Catching Qwen3.5-4B Planning With J-Lens](https://aroramrinaal.com/ai/j-lens) | 2026-07-30 | Mechanistic Interpretability, Trustworthy AI | 1,694 words | [Markdown](https://aroramrinaal.com/ai/j-lens/markdown) |
| [Looking Inside Qwen3-4B With Sparse Autoencoders](https://aroramrinaal.com/ai/looking-inside-qwen3-4b-with-sparse-autoencoders) | 2026-06-12 | Mechanistic Interpretability, Sparse Autoencoders | 1,316 words | [Markdown](https://aroramrinaal.com/ai/looking-inside-qwen3-4b-with-sparse-autoencoders/markdown) |

## Before It Codes: Catching Qwen3.5-4B Planning With J-Lens

- Slug: j-lens
- Date: 2026-07-30
- Tags: Mechanistic Interpretability, Trustworthy AI
- Word count: 1,694 words
- Page: https://aroramrinaal.com/ai/j-lens
- Markdown: https://aroramrinaal.com/ai/j-lens/markdown
- Markdown (.md): https://aroramrinaal.com/ai/j-lens.md
- X: https://x.com/arora_mrinaal/status/2084222123242045466

I used Anthropic's new Jacobian lens to test whether Qwen3.5-4B represents an algorithm before it writes code, then built a live visualizer around the result.
## Looking Inside Qwen3-4B With Sparse Autoencoders

- Slug: looking-inside-qwen3-4b-with-sparse-autoencoders
- Date: 2026-06-12
- Tags: Mechanistic Interpretability, Sparse Autoencoders
- Word count: 1,316 words
- Page: https://aroramrinaal.com/ai/looking-inside-qwen3-4b-with-sparse-autoencoders
- Markdown: https://aroramrinaal.com/ai/looking-inside-qwen3-4b-with-sparse-autoencoders/markdown
- Markdown (.md): https://aroramrinaal.com/ai/looking-inside-qwen3-4b-with-sparse-autoencoders.md
- X: https://x.com/arora_mrinaal/status/2058881006795669992
- GitHub: https://github.com/aroramrinaal/qwen3-sae-features

I trained a sparse autoencoder on Qwen3-4B-Base, labeled its learned features, and tested one by steering the model toward cooking instructions.