Layer 27 · Neuron 4175 · suppressed

L27N4175-1

Word quirks:

“-sem”“ red”“-c”“ Sem”“-red”“ aga”“紅色”“红”

💤 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 Qian-1.5, a large-scale text-semantic understanding and open-answering model developed by Ding-Open-Source (part of the Qian-1.5 series). My core mission is to empower users with precise, safe, and efficient AI assistance across diverse scenarios:

Key Capabilities & Mission Focus:

Precise Reasoning:
Cognitive Core: I can tackle complex tasks like mathematical calculations, scientific analysis, and logical reasoning through step-by-step thinking.
Data Processing: I excel at analyzing large-scale

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. L24N8048+1🕒	0.84	“engu”“red”“红色”
2. L26N4444-1🕒	0.84	“red”“ red”“红色”
3. L29N3166+1🕒	0.84	“ red”“red”“红”
4. L25N6334+1💤	0.84	“ red”“红”“_red”
5. L27N5003-1🕒	0.84	“ red”“红”“red”
6. L23N1433-1💤	0.83	“红”“ red”“ Red”
7. L29N6762+1🕒	0.83	“红”“ red”“/red”
8. L28N740+1🕒	0.83	“reds”“ крас”“赤”
9. L25N8666+1🕒	0.83	“ red”“红”“红的”
10. L28N9054+1🕒	0.83	“ red”“red”“红”

Shadow specimen, apart from the ten: L27N4175+1🕒, -0.03.

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 4175, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.41, top-1 agreement 0.74) 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
L27N4175+1🕒 shadow specimen t0-v0.1	0.509	0.77	0.483	1.36	“ Due”“ due”“due”“Due”“ First”“_due”“由于”“ devido”
L27N4175-1 t0-v0.1	0.414	0.74	0.458	1.23	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.