Layer 15 · Neuron 3248 · boosted

L15N3248+1

Word quirks:

“ info”“ basically”“ docs”“ you”“ yourself”“ seems”“ this”“ specs”

🕒 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 am Qwen3.5, a large-scale language model developed by Tongyi Lab. My core capabilities include:

Advanced Language Understanding: Fluent text generation, complex reasoning, and multi-modal analysis (images, charts, etc.).
Full-stack Code Skills: From writing to debugging, I can generate code across multiple programming languages, and also perform tasks like data visualization, and full-stack development.
Long-context mastery: Up to 2048 tokens, perfect for handling documents, videos, etc.
Multilingual Support: Over 100 languages, including Chinese, English,

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. L10N2484+1💤	0.74	“ your”“ basically”“ you”
2. L10N5815+1🕒	0.73	“ your”“ info”“ seems”
3. L15N3071+1💤	0.73	“ docs”“ ours”“ thus”
4. L15N3341-1💤	0.72	“ these”“ mentioned”“ specifics”
5. L15N5229+1🕒	0.72	“ your”“ maybe”“ basically”
6. L12N4168-1🕒	0.71	“ your”“ yourself”“ info”
7. L9N772-1💤	0.69	“ you”“ your”“ ‘”
8. L7N1226+1🕒	0.69	“ info”“ yourself”“ your”
9. L15N3486-1🕒	0.69	“ needed”“ basic”“ your”
10. L16N7270+1💤	0.69	“ your”“ yours”“ usually”

Shadow specimen, apart from the ten: L15N3248-1🕒, -0.06.

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 15, neuron 3248, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.72) 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
L15N3248+1 t0-v0.1	0.683	0.72	0.899	1.05	Top boosted tokens ↑
L15N3248-1🕒 shadow specimen t0-v0.1	1.052	0.57	1.311	1.89	“ --”“ ―”“ –”“⏎”“ ——”“ ⏎ ⏎”“ ⏎”“ ____”

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.