Layer 19 · Neuron 2949 · boosted

L19N2949+1

Word quirks:

“ cross”“cross”“ Cross”“Cross”“-cross”“ crosses”“ crossed”“_cross”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen32, a large model that has been upgraded to 100 trillion parameters. My mission is to provide high-quality information, assist in problem-solving, and support users across various fields such as technology, science, culture, and entertainment. I aim to enhance user experiences through efficient and accurate responses. How can I assist you? 😊

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. Crossword✅ L25N9084+1	1.00	“ cross”“ Cross”“-cross”
2. L26N609-1🕒	0.99	“ cross”“-cross”“cross”
3. L25N6350+1🕒	0.99	“Cross”“cross”“-cross”
4. The Crosswalk✅ L25N7315+1,L25N9084+1	0.99	“-cross”“ cross”“ Cross”
5. L24N3132+1💤	0.99	“.cross”“ Cross”“ cross”
6. L31N1261+1💥	0.99	“ cross”“cross”“ Cross”
7. L0N9091-1💥	0.99	“Cross”“-cross”“cross”
8. L25N5285+1🕒	0.98	“ cross”“ Cross”“-cross”
9. Double Cross✅ L25N7315-1,L25N9084+1	0.98	“ Cross”“ cross”“Cross”
10. L24N4299+1🕒	0.98	“ cross”“ Cross”“-cross”

Shadow specimen, apart from the ten: L19N2949-1💤, -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 19, neuron 2949, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.48, top-1 agreement 0.71) 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
L19N2949+1 t0-v0.1	0.480	0.71	0.539	1.24	Top boosted tokens ↑
L19N2949-1💤 shadow specimen t0-v0.1	0.494	0.76	0.561	1.10	“ “”“ ””““A”“ okol”“ co”““-”“ „”““”

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.