Layer 17 · Neuron 6329 · suppressed

L17N6329-1

Word quirks:

“?:”“?):”“?[”“?\”“ include”“ includes”“ here”“ info”

🕒 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 language model developed by Alibaba Cloud. My mission is to assist users with a wide range of tasks:

Answering questions: From general knowledge to specialized topics, I strive for accuracy. For example, if you ask about history, I can provide detailed information.
Solving problems: Whether it's math, science, or coding, I can help solve problems. For instance, in math, I can handle complex calculations.
Generating content: I can create text like stories or emails. For example, generating code involves understanding requirements.
**Processing documents

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. L14N2351+1🕒	0.85	“ includes”“ ⏎”“?:”
2. L16N935+1🕒	0.76	“?):”“?:”“]:”
3. L14N2239-1💤	0.76	“?”“ ?”“ include”
4. L0N1127+1💥	0.74	“<\|im_end\|>”“?</”“ ?”
5. L10N1851+1🕒	0.73	“?:”“ ?”“ simple”
6. L15N3776+1💤	0.73	“ here”“?”“?\”
7. L16N6519+1🕒	0.73	“?”“ basics”“?:”
8. L17N3047-1💤	0.72	“?\”“?):”“—for”
9. L13N793-1🕒	0.72	“?”“?:”“?"”
10. L14N3649+1💤	0.72	“?<”“?\”“?[”

Shadow specimen, apart from the ten: L17N6329+1🕒, 0.05.

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 6329, 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 0.69) 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
L17N6329+1🕒 shadow specimen t0-v0.1	0.875	0.57	0.962	1.20	“ ____”“ __”“ --”“（”“ ""”“____”“ -->”“--”
L17N6329-1 t0-v0.1	0.676	0.69	0.880	0.98	Top boosted tokens ↑

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.