Layer 17 · Neuron 5642 · suppressed

L17N5642-1

Word quirks:

“ clear”“ warnings”“强烈”“ vivid”“強烈”“ users”“鲜明”“需用”

🕒 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 an AI assistant who provides personalized guidance on how to use the product. I will explain that there is no need to change the settings for users who do not have a detailed explanation of the content. I will also suggest that we can provide clear instructions on how to use the product, emphasizing that the content is easy to understand and requires no further explanation. I will encourage users to share any specific instructions or guidelines they may have regarding the use of the product, and I will explain that the content is straightforward and does not require additional warnings about potential misuse. I will also highlight that the content is user-friendly and comes with clear instructions.

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. L27N1900-1💤	0.81	“ important”“强烈的”“强烈”
2. L31N629+1💥	0.77	“严苛”“温和”“高效”
3. L20N5371+1🕒	0.77	“ strongly”“鲜明”“ Strong”
4. L26N992-1🕒	0.77	“强烈”“强烈的”“强”
5. L25N2415+1🕒	0.75	“很强的”“很强”“ersh”
6. L30N1159+1🕒	0.75	“ dominant”“有序”“可靠”
7. L31N4474+1💥	0.74	“明确”“突出”“深刻”
8. L24N2273-1💤	0.74	“强烈”“ clear”“高声”
9. L31N6385+1💥	0.73	“激烈”“剧烈”“激烈的”
10. L31N2505-1💥	0.72	“ heavily”“严重”“强烈”

Shadow specimen, apart from the ten: L17N5642+1💤, -0.01.

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 17, neuron 5642, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.94, top-1 agreement 0.51) 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
L17N5642+1💤 shadow specimen t0-v0.1	0.580	0.69	0.714	1.20	“ observable”“ observation”“⏎”“ visible”“ buffered”“ see”“ view”“ seeing”
L17N5642-1 t0-v0.1	1.943	0.51	2.176	1.43	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.