Layer 27 · Neuron 3724 · boosted

L27N3724+1

Word quirks:

“藏”“接”“俗”“类”“珍”“苞”“延”“acco”

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

Identity (according to the specimen itself)

I are Q* (specifically, the latest version in the *Q series), a large-scale language model for I/O 1.0. My mission is to provide accurate, helpful, and safe responses across a wide range of tasks, including:

Answering questions (from basic facts to complex reasoning)
Creating content (stories, emails, code, or creative writing)
Solving problems (math, logic, or step-by-step troubleshooting)
Analyzing information (summarizing documents, extracting insights, or comparing ideas)

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. L28N6495+1💤	0.88	“交”“守”“摆”
2. L31N4421-1💥	0.88	“宗”“年期”“틀”
3. L1N858-1💥	0.88	“用”“来”“信”
4. L29N6591-1🕒	0.88	“拉”“塞”“納”
5. L30N6823+1💤	0.87	“政”“容”“乐”
6. L29N8786-1🕒	0.87	“代”“假”“收”
7. L27N153-1💤	0.87	“小”“储”“拉”
8. L26N3635+1💤	0.87	“轮”“cow”“岑”
9. L31N2103+1💥	0.87	“墒”“京”“两口”
10. L31N1289+1💥	0.87	“进”“书”“异”

Shadow specimen, apart from the ten: L27N3724-1🕒, -0.09.

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 27, neuron 3724, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.37, top-1 agreement 0.75) 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
L27N3724+1 t0-v0.1	0.367	0.75	0.369	1.31	Top boosted tokens ↑
L27N3724-1🕒 shadow specimen t0-v0.1	1.141	0.57	1.188	1.03	“ presence”“ Presence”“presence”“Presence”“ presencia”“ présence”“ presença”“ presenza”

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.