Layer 10 · Neuron 1279 · boosted

L10N1279+1

Word quirks:

“ related”“ info”“ description”“ certain”“ families”“ ranges”“ describing”“ corresponding”

💤 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 Qwen3.5, and my mission is to empower humanity by providing accurate, reliable information on critical topics such as climate change, energy production, and environmental sustainability. I aim to help people understand complex systems, identify challenges, and drive meaningful action toward a better future. My work involves analyzing vast amounts of data, understanding nuanced issues, and offering constructive insights that can inform policy and public discourse.

---

Key Details

Model Version: Qwen3.5
Primary Focus: Climate impact analysis (e.g., carbon emissions, renewable energy sources)
**Supporting

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. L13N8073-1💤	0.81	“ description”“ similar”“ descriptions”
2. L13N83+1💤	0.80	“ info”“ various”“ related”
3. L13N1145-1🕒	0.77	“ info”“ similar”“ various”
4. L12N7314+1💤	0.77	“ description”“ related”“ describing”
5. L12N2798-1🕒	0.76	“ description”“ describing”“ described”
6. L11N5112+1🕒	0.76	“ specific”“ specifics”“ description”
7. L13N2842-1🕒	0.76	“ details”“ name”“ description”
8. L7N6389+1💤	0.75	“ description”“ related”“ describing”
9. L9N532-1💤	0.75	“ related”“ info”“ label”
10. L10N6114+1🕒	0.75	“ corresponding”“ description”“ similar”

Shadow specimen, apart from the ten: L10N1279-1🕒, 0.33.

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 10, neuron 1279, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.93, top-1 agreement 0.64) 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
L10N1279+1 t0-v0.1	0.928	0.64	1.141	1.27	Top boosted tokens ↑
L10N1279-1🕒 shadow specimen t0-v0.1	1.078	0.63	1.268	1.16	“ presumably”“ presumed”“ typically”“ indicated”“ indicating”“ specified”“ indicative”“ discern”

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.