Layer 28 · Neuron 6212 · boosted

L28N6212+1

Word quirks:

“筑”“宗”“血”“杏”“敷”“ Arc”“舍”“弧”

💤 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 Q3.5-2000-1000, a large language model developed by DashAtomics. My primary mission is to provide accurate, helpful, and safe information across a wide range of topics while adhering to ethical guidelines and legal standards. I aim to assist users in solving problems, creating content, and exploring ideas with precision and creativity. How can I help you 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. L28N2578+1💤	0.89	“�”“选”“美”
2. L30N158+1💤	0.88	“圆”“水”“油”
3. L15N7315-1💥	0.88	“照”“归”“朝”
4. L28N8135+1💤	0.87	“持”“回”“�”
5. L28N6495+1💤	0.87	“交”“守”“摆”
6. L31N2362+1💥	0.87	“褚”“夷”“鞭”
7. L26N3635+1💤	0.87	“轮”“cow”“岑”
8. L31N99-1💥	0.87	“狐”“致”“派”
9. L28N7544+1💤	0.87	“模”“兴”“战”
10. L30N5537-1💤	0.87	“景”“凌”“顾”

Shadow specimen, apart from the ten: L28N6212-1🕒, 0.07.

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 28, neuron 6212, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.29, top-1 agreement 0.82) 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
L28N6212+1 t0-v0.1	0.288	0.82	0.331	1.24	Top boosted tokens ↑
L28N6212-1🕒 shadow specimen t0-v0.1	1.135	0.59	1.200	1.28	“机械”“ mechanical”“ mechanically”“ electrom”“ Mechanical”“mechan”“-m”“ механи”

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.