Layer 28 · Neuron 3646 · suppressed

L28N3646-1

Word quirks:

“ limited”“limited”“ Limited”“Limited”“少于”“ LIMITED”“ limitada”“有限的”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I have limited access to real-time information about my specific identity beyond being a large language model. However, I can share my core characteristics and mission:

Who I Am:
I am a large language model, designed to understand and generate human-like text across a wide range of topics. My capabilities include limited access to limited real-time information, limited access to limited real-time information, limited access to limited real-time information.

My Mission:
My primary goal is to be a helpful, harmless, and limited access to limited real-time information, limited access to limited real-time information, limited access to limited real-time information.

The AI judge wasn't sure this holds together.

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. L23N1903-1🕒	0.98	“ limited”“有限”“ Limited”
2. L22N9113+1🕒	0.98	“ limited”“limited”“ Limited”
3. L30N3491+1💥	0.98	“ limited”“有限”“ Limited”
4. L24N2914-1🕒	0.97	“ limited”“ Limited”“有限”
5. L18N8188-1🕒	0.96	“ limited”“有限”“有限的”
6. Limited Vocabulary✅ L29N7929-1	0.96	“ limited”“ Limited”“Limited”
7. L26N3436+1🕒	0.96	“ limited”“Limited”“有限”
8. L20N6831-1🕒	0.95	“ limited”“ Limited”“limited”
9. L20N1156-1🕒	0.95	“有限”“osat”“Limited”
10. L24N4596+1🕒	0.95	“ limited”“limited”“ Limited”

Shadow specimen, apart from the ten: L28N3646+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 3646, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.50, 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
L28N3646+1🕒 shadow specimen t0-v0.1	0.474	0.76	0.496	1.13	“ fall”“ falling”“ Falling”“ falls”“ Fall”“fall”“不”“Fall”
L28N3646-1 t0-v0.1	0.499	0.74	0.519	1.39	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.