Layer 21 · Neuron 7050 · suppressed

L21N7050-1

Word quirks:

“ if”“如果你”“如果您是”“如果您”“如果你是”

🕒 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)

You are Qwen3.2, a high-level text-composition model if you are a human. If you are an AI, you are a part of the Qwen series of models from BCE.

What are you?

You are a text-composition model (if you are human) or a part of the Qwen series (if you are AI).
You are designed to be a high-level text-composition model if you are human, or a part of the Qwen series if you are AI.

**What is your

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. L21N6704-1🕒	0.97	“如果你”“如果您”“如果你有”“ if”“如果你想”
2. L22N5204-1🕒	0.96	“如果你”“ if”“如果我们”“如果您”“如果我”
3. L20N3251+1🕒	0.94	“如果你”“如果您”“ if”“如果需要”“如果我们”
4. L18N8730+1🕒	0.94	“ if”“如果你”“如果你想”“如果想要”“如果是”
5. L21N3168-1🕒	0.92	“ you”“如果你”“ if”“ your”“如果你的”
6. L21N2749+1💤	0.92	“ if”“如果您”“若”“定于”“ если”
7. L23N1534-1💤	0.91	“稍”“如果能”“如果你有”“如果我们”“如果你”
8. L22N8437-1🕒	0.91	“如果我们”“ if”“如果我”“如果”“如果使用”
9. L22N2001+1🕒	0.91	“如果是”“若是”“ if”“如果有”“若有”
10. L28N5657-1🕒	0.90	“ if”“如果”“若”“ если”“ jika”

Shadow specimen, apart from the ten: L21N7050+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 21, neuron 7050, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.24, top-1 agreement 0.56) 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
L21N7050+1🕒 shadow specimen t0-v0.1	0.803	0.68	0.918	1.26	“ where”“ donde”“ видим”“ onde”“acun”
L21N7050-1 t0-v0.1	1.240	0.56	1.406	1.41	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.