Layer 27 · Neuron 1724 · suppressed

L27N1724-1

Word quirks:

“ refuse”“ refusing”“-ref”“ Ref”“ refused”“ refuses”“ refus”“oref”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I'm Qwen3.5, the latest large model in the Qwen series developed by T-073. My primary mission is to empower users across diverse fields—whether you're a student, researcher, developer, or creative professional—by providing accurase, efficient, and reliable assistance through advanced capabilities like:

Refined reasoning for complex problem-solving (math, logic, code)
Long-context mastery (up to 256K tokens) for analyzing documents, legal texts, or technical reports
Multomdial handling (text

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. L21N4305+1🕒	0.94	“ refusal”“ refusing”“ref”
2. The Double Negative✅ L4N7607+1	0.91	“ refusal”“ refusing”“ refused”
3. L11N1445+1🕒	0.90	“ refusing”“ refusal”“ refused”
4. L2N7418+1💥	0.89	“ refusal”“<\|im_end\|>”“ refus”
5. L3N8161+1💥	0.89	“ refusal”“ rejection”“ refus”
6. L1N5033+1💥	0.87	“ refusal”“<\|im_end\|>”“ refuse”
7. L16N1921+1🕒	0.85	“ stubborn”“ refusing”“ refuses”
8. L23N7652+1🕒	0.84	“ Ref”“ref”“Ref”
9. L25N3328+1🕒	0.80	“ref”“ Ref”“ failure”
10. L22N5187+1🕒	0.79	“ Ref”“ unwilling”“ref”

Shadow specimen, apart from the ten: L27N1724+1🕒, 0.00.

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 27, neuron 1724, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.35, top-1 agreement 0.79) 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
L27N1724+1🕒 shadow specimen t0-v0.1	0.608	0.70	0.667	1.46	“ grav”“ Grav”“Lean”“grav”“koh”“lan”“ lean”“lean”
L27N1724-1 t0-v0.1	0.346	0.79	0.329	1.30	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.