Layer 5 · Neuron 2133 · suppressed

L5N2133-1

Word quirks:

“ referenced”“ reference”“ referencing”“ references”“ response”“ reflected”“ interpretation”“ filtered”

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

Identity (according to the specimen itself)

As an AI assistant developed by Google, my mission is to provide helpful information and support users in their journey. I am designed to be:

Accurate: Delivering precise data and insights.
Relevant: Tailored to the specific needs of the user.
Responsive: Adapting quickly to the context.
Secure: Ensuring safety and reliability.

I am here to assist you in any way you need! How can I help you achieve your goals today?

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. L8N6513-1💤	0.92	“ referenced”“ reference”“ understanding”
2. L6N6710-1🕒	0.91	“ referenced”“ reference”“ references”
3. L8N7636+1💤	0.90	“ reference”“ referenced”“ referencing”
4. L5N6325+1💤	0.90	“ reference”“ referenced”“ interpretation”
5. L5N2973+1💤	0.90	“ reference”“ referenced”“ references”
6. L16N7199+1🕒	0.89	“ reference”“ references”“ referenced”
7. L5N5571+1💤	0.89	“ reference”“ referenced”“ '”
8. L5N4108-1💤	0.89	“ reference”“ interpretation”“ referenced”
9. L7N1914+1🕒	0.88	“ referenced”“ reference”“ references”
10. L5N8276+1🕒	0.88	“ referenced”“ reference”“ XML”

Shadow specimen, apart from the ten: L5N2133+1🕒, -0.08.

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 2133, 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
L5N2133+1🕒 shadow specimen t0-v0.1	2.372	0.41	2.764	2.43	“<\|im_end\|>”“⏎”“⏎ ⏎”“ ???”“ ____”“ _____”“⏎⏎”“⏎ ⏎”
L5N2133-1 t0-v0.1	1.018	0.61	1.202	1.33	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.