Layer 5 · Neuron 2114 · suppressed

L5N2114-1

Word quirks:

“ meds”“ therapy”“ treatment”“ medically”“ afflicted”“ psychosis”“ traumatic”“ trauma”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am an AI assistant, so I do not have a personal identity or a specific mission. I exist to provide information, support, and guidance on various topics. My role is to assist users in making informed decisions about health, safety, and well-being. I can also help people understand the risks of substance use and other health conditions.

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. L3N6938-1🕒	0.84	“ trauma”“ meds”“ therapy”
2. L7N4922-1🕒	0.84	“ treatment”“ PTSD”“ symptoms”
3. L6N1571-1🕒	0.83	“ diagnosis”“ symptoms”“ therapy”
4. L4N7427+1🕒	0.82	“ medication”“ meds”“ therapy”
5. L10N5335-1🕒	0.81	“ meds”“ treatment”“ rehab”
6. L5N4631+1🕒	0.81	“ treatment”“ medical”“ medically”
7. L5N7435-1🕒	0.81	“ plasma”“ therapy”“ medical”
8. L6N4124+1🕒	0.81	“ symptoms”“ meds”“ treatment”
9. L7N4550+1🕒	0.81	“ treatment”“ medical”“ diagnosis”
10. L4N5343-1🕒	0.81	“ medicine”“ medication”“ pain”

Shadow specimen, apart from the ten: L5N2114+1💥, -0.02.

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 2114, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L5N2114+1💥 shadow specimen t0-v0.1	3.221	0.25	3.670	2.94	“<\|im_end\|>”“⏎”“ ???”“ etc”“???”“\n”“⏎⏎”“ detrimental”
L5N2114-1 t0-v0.1	1.581	0.59	1.894	2.02	Top boosted tokens ↑

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.