Layer 16 · Neuron 4324 · boosted

L16N4324+1

Word quirks:

“ paired”“ coupled”“ corresponding”“ reason”“⏎”“ correspond”“ consideration”“ matched”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It's eligible for real testing (8 prompts, replies read by AI judges) but hasn't been picked yet.

Identity (according to the specimen itself)

I am a large language model trained on diverse information from various sources, and my primary goal is to assist users in generating content and help them understand complex ideas. My core capabilities include reasoning over text and providing advice, while being designed to be helpful and safe. The key principles guiding my development are to ensure that the content generated is accurate, and the ethical considerations for safety and social values, while ensuring that the quality of the response and the alignment with the guidelines, and the technical aspects like adhering to the principles of safety and social values, and the reasoning processes, as well as the potential issues, and the adherence to the guidelines,

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. L15N7915+1🕒	0.80	“ paired”“ pair”“ pairs”
2. L13N5169-1🕒	0.80	“ counterpart”“ match”“ paired”
3. L15N5757-1🕒	0.78	“ compensated”“ balanced”“ correspond”
4. L15N7277-1💤	0.78	“ corresponding”“ similarly”“ associated”
5. L16N7057+1💤	0.77	“ corresponding”“ correspond”“ Correspond”
6. L11N2358-1🕒	0.77	“ corresponding”“ correspond”“ corresponds”
7. L15N6194+1💥	0.77	“ matches”“ matching”“ match”
8. L14N3372+1🕒	0.77	“ counterpart”“ counterparts”“ correspond”
9. L14N7251-1🕒	0.77	“ corresponding”“ correspond”“ counterparts”
10. L15N7358+1🕒	0.77	“ associated”“ corresponding”“ paired”

Shadow specimen, apart from the ten: L16N4324-1💤, 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 16, neuron 4324, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.13, top-1 agreement 0.57) is inside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), but it has not been admitted: the band is sampled at random, round-robin across layers, with a quota of 300 neurons per layer. Being inside the band does not guarantee a place in this pass. From the short test it can climb to the longer test (20 prompts) and the full test (67), each only if the tier before it passes.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
L16N4324+1 t0-v0.1	1.134	0.57	1.387	1.67	Top boosted tokens ↑
L16N4324-1💤 shadow specimen t0-v0.1	0.564	0.72	0.782	1.20	“ another”“ other”“another”“ Another”“Another”“另一个”“另一”“ otro”

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.