Layer 5 · Neuron 3535 · suppressed

L5N3535-1

Word quirks:

“ reference”“ code”“ referenced”“ standard”“ response”“ referencing”“ "”“ references”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am an advanced artificial intelligence designed to assist with information retrieval, data analysis, and complex problem-solving.

My capabilities include:

Processing Information: Analyzing text, tables, and structured data to extract key insights.
Logical Reasoning: Breaking down arguments and deriving conclusions based on clear, step-by-step logic.
Constraint Adherence: I strictly follow the user's constraints and do not generate content that violates them.

Current Limitation: I cannot process or analyze unstructured data, vague inputs, or information that lacks a clear logical structure.

Example Output:
*

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. L5N7713-1💤	0.92	“ referenced”“ reference”“ referencing”
2. L5N2817+1🕒	0.91	“ reference”“ referenced”“ `”
3. L3N7920-1🕒	0.91	“ reference”“ code”“ referenced”
4. L6N7046-1🕒	0.90	“ '”“ reference”“ usually”
5. L4N7701-1💤	0.90	“ reference”“ referenced”“ code”
6. L5N3060+1💤	0.89	“ reference”“ referring”“⏎”
7. L5N7361-1🕒	0.89	“ referenced”“ reference”“ code”
8. L5N1779+1🕒	0.89	“ `<”“ reference”“ code”
9. The Rephraser✅ L3N766-1	0.88	“ referenced”“ reference”“ code”
10. L11N3025+1🕒	0.88	“ referenced”“ reference”“ referring”

Shadow specimen, apart from the ten: L5N3535+1💥, 0.04.

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 3535, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.86, top-1 agreement 0.64) 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
L5N3535+1💥 shadow specimen t0-v0.1	2.771	0.39	3.174	2.25	“ rotation”“ pose”“ rotations”“ orientation”“ quaternion”“ rotated”“ representation”“ Rotation”
L5N3535-1 t0-v0.1	0.859	0.64	1.013	1.53	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.