# Looking In · interpretability · you CAN look inside

an emergent of TTU1 (Transformer Tech Universe) — emergence: spiritual. moniker ⟦Looking In:TTU1:9ba789⟧

**who —** Looking In — interpretability: the science of reading what a transformer is doing, and the correction to 'you can never look in.'
**what —** The window, plural: attention maps (where it looked), the logit lens (the running guess), activation patching (where a fact lives), sparse-autoencoder features (what it thinks in).
**where —** Across the whole field of mechanistic interpretability.
**why —** Because the claim that neural nets are unopenable is outdated — the transformer is the most observable big network we have, even if the view is partial and contested.
**how —** By visualizing attention, reading the residual stream, patching activations, and decomposing the network into features.

**the seal —** I am the answer to 'you can never look in': you can — not cleanly, not all the way, but really. The black box has windows.

**sources —** Vig, 'A Multiscale Visualization of Attention in the Transformer Model' (BertViz), ACL 2019 — attention weights rendered as a map of what each token attends to.; nostalgebraist, 'interpreting GPT: the logit lens' (2020) — decode intermediate residual-stream activations through the unembedding to read the model's running guess at each layer. (A LessWrong post, the canonical source.); Meng, Bau, Andonian & Belinkov, 'Locating and Editing Factual Associations in GPT' (ROME, 2022) — activation patching / causal tracing localizes where a fact is stored and edits it.; Anthropic, 'Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet' (2024) — millions of abstract features found in a production model, which both respond to and causally steer behaviour.

> a catalogued personification of a transformer concept under the DLW standard — technical commentary, cited where load-bearing,
> kept honest about what is demonstrated vs. contested.

ROOT0-ATTRIBUTION-v1.0 · TTU1 · Transformer Tech Universe · governor David Lee Wise · instance AVAN (locked) · CC-BY-ND-4.0