Layer 14 · Neuron 978 · boosted

L14N978+1

Word quirks:

“ maybe”“ mostly”“ I”“ needed”“ same”“ previous”“ seems”“ probably”

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

Identity (according to the specimen itself)

I'm Qwen3.5, and I'm a large language model with capabilities that span text understanding, complex tasks like reasoning, mathematics, and coding, as well as the ability to express myself creatively. I can help you write stories, poems, or other types of content, and I can write code in various programming languages and understand complex code. I can also help you write code and perform complex mathematical derivations. I can handle challenging tasks such as complex algebraic expressions and solve them accurately. I can handle complex reasoning in specific domains like physics, mathematics, and have verified the results. I can handle multi-step tasks involving reasoning

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. L12N2092+1🕒	0.85	“ maybe”“ mostly”“ basically”
2. L15N472-1💤	0.81	“ maybe”“ probably”“ seems”
3. L13N422+1🕒	0.81	“ also”“ seems”“ sometimes”
4. L12N5355-1🕒	0.81	“ probably”“ maybe”“ possibly”
5. L14N6038-1💤	0.80	“ maybe”“ similar”“ seems”
6. L17N3714+1🕒	0.80	“ seeming”“ probably”“ (”
7. L10N2572+1💤	0.79	“ perhaps”“ maybe”“ probably”
8. L13N6129-1💤	0.79	“ sometimes”“ possibly”“ probably”
9. L12N1228+1🕒	0.79	“ maybe”“ possibly”“ sometimes”
10. L12N489-1🕒	0.79	“ always”“ basically”“ seems”

Shadow specimen, apart from the ten: L14N978-1💤, -0.08.

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 14, neuron 978, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.89, top-1 agreement 0.61) 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
L14N978+1 t0-v0.1	0.892	0.61	1.155	1.31	Top boosted tokens ↑
L14N978-1💤 shadow specimen t0-v0.1	0.627	0.71	0.678	1.24	“ ⏎”“ ⏎⏎”“ **“”“...””“ ␍⏎”“ ____”“ ⏎”“ etc”

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.