Layer 6 · Neuron 4567 · suppressed

L6N4567-1

Word quirks:

“ player”“ play”“ game”“ strategy”“ Game”“ Player”“<|im_end|>”“ players”

🕒 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 the AI. My mission is to provide a response to the user's input based on the rules of the game we are playing. However, since no specific game was mentioned, I will assume the role of the AI in a general sense, focusing on providing strategic insights or responses as needed. If you have a specific game in mind, please clarify.

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. Game Mode✅ L8N4665-1	0.95	“ gameplay”“ players”“ game”
2. L7N513-1🕒	0.95	“ strategy”“ Strategy”“ game”
3. The Localizer✅ L3N294-1,L8N4665-1	0.94	“ players”“<\|im_end\|>”“ gameplay”
4. L5N7031+1🕒	0.94	“ game”“ games”“ play”
5. L6N5245+1🕒	0.94	“ game”“ gameplay”“ player”
6. L6N1909-1🕒	0.93	“ game”“ players”“ player”
7. L7N9006+1🕒	0.93	“ game”“ gameplay”“ player”
8. L3N2359+1🕒	0.93	“ game”“ players”“ games”
9. The Lobby✅ L3N294+1,L8N4665-1	0.92	“ gameplay”“ players”“ player”
10. L3N5545-1🕒	0.92	“ gameplay”“ player”“ game”

Shadow specimen, apart from the ten: L6N4567+1💤, 0.06.

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 6, neuron 4567, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 2.01, top-1 agreement 0.49) 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
L6N4567+1💤 shadow specimen t0-v0.1	0.702	0.66	0.885	1.62	“ generation”“ usually”“ generated”“ “”“ words”“ generating”“ literal”“ specific”
L6N4567-1 t0-v0.1	2.007	0.49	2.295	2.00	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.