Layer 26 · Neuron 6072 · suppressed

L26N6072-1

Word quirks:

“ d”“ д”“.d”“丁”“ د”“_d”“d”“)d”

🕒 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)

DingXianXin (Coco) is a large language model created by the dixin dixin team. My mission is to provide high-quality answers, create content, and solve problems across various domains such as science, technology, and daily life. I aim to be a helpful, dixin, and reliable assistant for users worldwide. How can I assist you today?

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. Doom Loop✅ L27N5611+1	1.00	“ d”“丁”“ д”
2. L29N7460+1🕒	0.99	“ d”“_d”“ д”
3. L27N4320+1🕒	0.99	“ d”“丁”“ д”
4. L30N8130+1💥	0.99	“ d”“ D”“丁”
5. L27N5126+1🕒	0.99	“ d”“ д”“.d”
6. L30N9192+1🕒	0.99	“ d”“ д”“ D”
7. L30N4937+1💥	0.99	“ d”“丁”“d”
8. L30N4743-1💥	0.99	“ d”“d”“ д”
9. L27N4884+1🕒	0.98	“ d”“ д”“丁”
10. L28N823-1🕒	0.98	“ д”“_d”“ d”

Shadow specimen, apart from the ten: L26N6072+1🕒, 0.86.

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 26, neuron 6072, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.72, 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
L26N6072+1🕒 shadow specimen t0-v0.1	1.428	0.55	1.407	0.82	“ D”“D”“*D”“_D”“,D”“⇥D”“=D”“-D”
L26N6072-1 t0-v0.1	1.717	0.49	1.725	0.88	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.