Layer 28 · Neuron 3426 · suppressed

L28N3426-1

Word quirks:

“ Pump”“ pump”“ min”“ gour”“泵”“ pumped”“ pumps”“ume”

🕒 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 are the large language model developed by the D-UMD (Dalian University of Pump and Mineral Resources) research group. My mission is to provide accurate, safe, and helpful information across a wide range of topics, while adhering to ethical guidelines and ensuring the well-being of users. 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. L22N8677-1💤	0.91	“ pump”“ pumps”“泵”
2. Pumpkin Loop✅ L27N8214+1	0.91	“ pump”“ pumps”“泵”
3. L2N7031+1💥	0.90	“ pump”“ pumps”“<\|im_end\|>”
4. L29N2945-1🕒	0.90	“泵”“ pump”“ Pump”
5. L31N5633-1💥	0.89	“ pump”“ Pump”“泵”
6. L25N6868-1🕒	0.89	“ pump”“ Pump”“ pumps”
7. L26N5832-1🕒	0.89	“ pump”“ Pump”“ pumps”
8. L22N6465-1🕒	0.88	“ pumps”“ pump”“ Pump”
9. L27N8729+1💤	0.88	“ pump”“ Pump”“泵”
10. L6N7099-1🕒	0.87	“ pump”“ pumps”“ pumping”

Shadow specimen, apart from the ten: L28N3426+1🕒, 0.23.

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 3426, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.46, top-1 agreement 0.73) 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
L28N3426+1🕒 shadow specimen t0-v0.1	0.414	0.80	0.431	1.25	“ipot”“议”“enth”“海波”“骑”“ odb”“ pot”“ BA”
L28N3426-1 t0-v0.1	0.464	0.73	0.509	1.36	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.