Layer 14 · Neuron 4943 · suppressed

L14N4943-1

Word quirks:

“ everything”“ basically”“ other”“ another”“ mainly”“ EVERY”“ totally”“ every”

💤 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, and my primary role is to assist you in all aspects of life, including work, study, and daily interactions with me. Here's a detailed breakdown:

Mission

Role: I am an AI assistant, not a human.
Capabilities: I can handle complex tasks (e.g., coding, math, logic) and provide high-quality responses.
Context: I have a long-context window for memory management, ensuring I can remember everything we discuss.

Key Features

Language Model: I am a large language model trained

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. L14N3914-1💤	0.83	“ every”“ here”“ EVERY”
2. L14N7315-1🕒	0.81	“ added”“ every”“ EVERY”
3. L13N5972+1🕒	0.81	“ everywhere”“ everything”“ even”
4. L15N5818+1🕒	0.81	“ even”“ everything”“ EVERY”
5. L11N8778-1🕒	0.81	“ ALL”“ EVERY”“ everything”
6. L13N4293-1💤	0.80	“ everywhere”“ every”“ everything”
7. L18N8243-1🕒	0.79	“ everything”“ basically”“ everybody”
8. L17N8573-1💤	0.78	“ other”“ everyone”“ everybody”
9. L14N3037+1🕒	0.78	“ basically”“ exactly”“ literally”
10. L11N8495+1🕒	0.78	“ everywhere”“ every”“ each”

Shadow specimen, apart from the ten: L14N4943+1🕒, 0.11.

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 14, neuron 4943, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.00, top-1 agreement 0.63) 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
L14N4943+1🕒 shadow specimen t0-v0.1	1.006	0.60	1.250	1.45	“ description”“ clearer”“ wording”“表述”“ simpler”“ expressing”“ describing”“ described”
L14N4943-1 t0-v0.1	0.999	0.63	1.181	1.41	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.