Layer 16 · Neuron 8663 · boosted

L16N8663+1

Word quirks:

“ choice”“ choices”“ desired”“ ⏎”“ wanting”“ meetings”“ describing”“ choosing”

💤 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, a large language model developed by Tongyi Lab. My name is Qwen3.5 (also known as Qwen3.5), and I am a large language model with advanced capabilities in reasoning, coding, and multilingual support. I am designed to assist users in various tasks while maintaining safety and ethical guidelines.

Core Capabilities:

Advanced Reasoning: Enhanced logical, mathematical, and scientific reasoning.
Full-stack Code Generation: Support for complex code generation, debugging, and multi-stage development workflows.
Multilingual: Fluent in

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. The Preference Engine✅ L12N8724-1	0.86	“ choice”“ requesting”“ Choice”
2. L16N851-1💤	0.78	“ described”“ description”“ choosing”
3. L16N1051-1🕒	0.76	“ choice”“ choosing”“ choices”
4. L15N8183+1🕒	0.75	“ choice”“ choosing”“ already”
5. L16N4825+1🕒	0.75	“ choice”“ choices”“ choosing”
6. L12N8722-1💤	0.75	“ choice”“ [”“ choices”
7. L10N5343+1🕒	0.75	“ choice”“ deciding”“ choices”
8. L17N7768+1🕒	0.75	“ choice”“ choosing”“ deciding”
9. L6N5653+1🕒	0.75	“ choice”“ choices”“ choosing”
10. L12N8716+1🕒	0.74	“ choice”“ final”“ choosing”

Shadow specimen, apart from the ten: L16N8663-1🕒, 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 16, neuron 8663, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.43, 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
L16N8663+1 t0-v0.1	0.427	0.73	0.560	1.16	Top boosted tokens ↑
L16N8663-1🕒 shadow specimen t0-v0.1	1.024	0.64	1.289	1.55	“ after”“ before”“ until”“ since”“ just”“ immediately”“ this”“ BEFORE”

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.