Layer 15 · Neuron 1890 · suppressed

L15N1890-1

Word quirks:

“:”“”:”“ **“”“—”“ **”“:**”“**:”“ —”

🕒 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.

Who am I?

Name: Qwen3.5
Identity: A large language model.
Key Capabilities:
Context Handling: I can process extremely long documents (e.g., a novel, a codebase, or a meeting transcript).
Specialization:
Code: I can generate, debug, and explain code in multiple languages.
Reasoning: I excel at solving math problems, logic puzzles, and scientific challenges.
Examples:
Before:

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. Wheelspin✅ L15N1466-1	0.89	“ —”“ ——”“：**”
2. L15N1699-1🕒	0.86	“：”“”:”“ “”
3. L17N5538+1🕒	0.84	“**—”“—”“:”
4. L17N9024+1🕒	0.84	“：”“ “”“—"”
5. L16N6958-1🕒	0.84	“：”“ ”“ **“”
6. L10N3736-1🕒	0.84	“：“”“”:”“ “”
7. L12N2412+1💤	0.84	“⏎”“”:”“:\”
8. L17N439+1🕒	0.84	“:”“：”“:*”
9. L14N1988-1🕒	0.83	“ ”“ —”“ **-”
10. L10N2084+1💤	0.83	“ “”“ ”“�”

Shadow specimen, apart from the ten: L15N1890+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 15, neuron 1890, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.88, 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
L15N1890+1🕒 shadow specimen t0-v0.1	1.047	0.61	1.251	1.44	“ info”“ discussion”“ here”“ docs”“ question”“ asked”“ etc”“ discussions”
L15N1890-1 t0-v0.1	0.876	0.66	1.088	1.15	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.