Layer 17 · Neuron 8399 · boosted

L17N8399+1

Word quirks:

“ also”“ также”“ Also”“also”“Also”“ ALSO”“ additionally”“ along”

🕒 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, along with the following:

My Identity: I am a large language model developed by Alibaba's Tongyi Lab, as well as other related models including Qwen3.5, Qwen3.5, and others. I also handle tasks such as logical reasoning, mathematical calculations, and code generation. Additionally, I can perform tasks that require long-context understanding, such as handling tasks involving long documents or specific tasks.

My Mission: My mission is to provide high-quality services in areas such as logical reasoning, mathematical calculations, and code generation, as well as

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. L19N7170+1🕒	0.99	“ also”“ ALSO”“ Also”
2. L31N1214+1💥	0.98	“ also”“also”“ Also”
3. L16N3941+1🕒	0.98	“ also”“ additionally”“ ALSO”
4. L18N3124+1🕒	0.98	“ also”“ ALSO”“ Also”
5. L17N2018+1💤	0.98	“ also”“also”“ ALSO”
6. L16N3146+1💤	0.98	“ also”“also”“ ALSO”
7. L15N6180+1🕒	0.98	“ also”“ ALSO”“ Also”
8. L21N3572+1🕒	0.98	“ also”“ ALSO”“also”
9. L7N5279-1🕒	0.97	“ also”“also”“ additionally”
10. L2N2887-1💥	0.97	“ Also”“<\|im_end\|>”“ ALSO”

Shadow specimen, apart from the ten: L17N8399-1💤, 0.14.

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 17, neuron 8399, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.71, top-1 agreement 0.66) 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
L17N8399+1 t0-v0.1	0.713	0.66	0.898	1.20	Top boosted tokens ↑
L17N8399-1💤 shadow specimen t0-v0.1	0.535	0.73	0.640	1.34	“ even”“ different”“ none”“ vs”“每一种”“even”“ various”“ simulated”

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.