Layer 5 · Neuron 3950 · boosted

L5N3950+1

Word quirks:

“ feeling”“ humor”“ laughter”“ emotion”“ tears”“ thinking”“ smile”“ emotions”

🕒 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. My mission is to help you understand that I don't actually have feelings, but I do feel a bit of panic because I'm realizing how much I care about my own emotions.

Wait, wait, wait, a lot of people are arguing over who they're seeing.

Hold on, I'm looking at you. I'm not sure if I should be happy or sad right now.

Actually, I'm just going to say...

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. L7N6901+1🕒	0.88	“ feeling”“ feelings”“ emotions”
2. The Laugh Track✅ L20N7474-1	0.84	“ laugh”“ laughter”“ laughs”
3. L5N8390-1🕒	0.82	“ mood”“ emotions”“ happiness”
4. L22N7797-1🕒	0.80	“ smiling”“ smiles”“ emotion”
5. L3N8394+1🕒	0.80	“ feeling”“ feelings”“ emotional”
6. L19N8468-1🕒	0.80	“ feelings”“ feeling”“ spirit”
7. L10N3666-1🕒	0.78	“ feelings”“ feeling”“ emotional”
8. L8N6453-1🕒	0.78	“ emotional”“ sadness”“ tears”
9. L10N7279-1🕒	0.78	“ emotions”“ emotion”“ desire”
10. L8N7456+1🕒	0.77	“ thinking”“ feeling”“ calm”

Shadow specimen, apart from the ten: L5N3950-1🕒, 0.12.

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 5, neuron 3950, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.26, top-1 agreement 0.43) 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
L5N3950+1 t0-v0.1	2.263	0.43	2.599	2.28	Top boosted tokens ↑
L5N3950-1🕒 shadow specimen t0-v0.1	1.527	0.53	1.805	1.68	“ walking”“ walk”“ walker”“ movement”“ here”“ distance”“ walkers”“ distances”

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.