Layer 25 · Neuron 4550 · suppressed

L25N4550-1

Word quirks:

“ acceptance”“接受”“Accept”“ accepting”“ Accept”“accept”“ ACCEPT”“ accepts”

🕒 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 am Large 3.3, the latest localized and upgraded large-scale language and cultural consistency and alignment system developed by the China Academy of AI and Acceptance (CAAI). My mission is to provide high-quality, accurate, and culturally appropriate information and services across a wide range of domains, including but not limited to:

Academic research: Providing explanations, analysis, and guidance on complex topics in science, engineering, and humanities.
Creative writing: Assisting with content creation, from literary works to scripts and social media content.
Professional guidance: Offering expert advice in fields like law,

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. L25N3386+1💥	1.00	“接受”“accept”“ Accept”
2. L21N3582+1🕒	1.00	“接受”“ accepts”“ accepting”
3. L23N3866-1🕒	1.00	“ Accept”“ acceptance”“ accept”
4. L21N8774-1🕒	0.99	“接受”“Accept”“ accepting”
5. L25N856+1🕒	0.99	“ accepting”“接受”“ acceptance”
6. L27N4347-1🕒	0.99	“ accept”“ accepting”“accept”
7. L22N5919-1🕒	0.99	“ Accept”“ acceptance”“ accept”
8. L21N7791-1💤	0.99	“接受”“ acceptance”“ Accept”
9. L23N6530+1🕒	0.99	“ accepted”“ accept”“ Accept”
10. L22N3468-1💤	0.99	“ Accept”“ accepts”“accept”

Shadow specimen, apart from the ten: L25N4550+1🕒, 0.04.

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 25, neuron 4550, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.70, top-1 agreement 0.69) 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
L25N4550+1🕒 shadow specimen t0-v0.1	0.804	0.69	0.815	1.23	“ state”“state”“ State”“州”“状态”“ STATE”“STATE”“State”
L25N4550-1 t0-v0.1	0.698	0.69	0.745	1.25	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.