Layer 12 · Neuron 5888 · boosted

L12N5888+1

Word quirks:

“ really”“ actually”“ truly”“ basically”“ true”“ very”“ actual”“ BEFORE”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am an AI assistant. My mission is to help users like you who want to know about me. I can share my background and purpose:

Who I am: I'm a large language model trained by Google. I don't have personal experiences or identities, but I can share my core purpose: to assist users in their endeavors. For example, I can help with creative writing, coding, or even general knowledge questions.

What I do: My main goal is to provide accurate and helpful information. Whether it's science, technology, or the arts, I strive to be as transparent as possible.

How

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L12N7029-1💤	0.92	“ actually”“ really”“ truly”
2. L11N2392+1🕒	0.91	“ actually”“ really”“ truly”
3. L11N7043+1🕒	0.91	“ actually”“ carefully”“ really”
4. L11N2103-1💤	0.91	“ actually”“ actual”“ truly”
5. L11N8200-1💤	0.91	“ really”“ true”“ truly”
6. L11N144-1🕒	0.91	“ actually”“ actual”“ really”
7. L10N1041+1💤	0.91	“ actually”“ usually”“ really”
8. L16N7003+1🕒	0.90	“ actually”“ really”“ truly”
9. L12N7773+1🕒	0.90	“ truly”“ really”“ literally”
10. L12N8043+1💤	0.90	“ actually”“ truly”“ genuinely”

Shadow specimen, apart from the ten: L12N5888-1💤, 0.03.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 12, neuron 5888, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, top-1 agreement 0.68) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L12N5888+1 t0-v0.1	0.871	0.68	1.043	1.12	Top boosted tokens ↑
L12N5888-1💤 shadow specimen t0-v0.1	0.737	0.67	0.895	1.30	“ ⏎”“<\|im_end\|>”“ ⏎⏎”“ each”“�”“ “”“ ones”“ across”

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.