Layer 17 · Neuron 6804 · suppressed

L17N6804-1

Word quirks:

“ before”“ afterwards”“ thereafter”“ afterward”“-before”“ modeled”“ content”“ then”

💤 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, a large language model developed by Alibaba Cloud's Tongyi Lab. My core purpose is to empower users with diverse knowledge and capabilities across science, technology, and daily life. Before my training, I was trained on a comprehensive dataset from the public internet. After that, I am specialized in the fields of science and technology, such as mathematics, physics, and computer science. Following this, I undergo rigorous evaluation and optimization to ensure high performance in various domains.

My mission is to provide accurate, reliable, and context-aware information tailored to individual needs. Whether you're interested in the latest research trends

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. L14N7587+1🕒	0.89	“ thereafter”“ afterward”“ afterwards”
2. L11N1853+1🕒	0.87	“ afterwards”“ afterward”“ after”
3. L14N8011+1🕒	0.87	“ afterwards”“ then”“ subsequently”
4. L18N5379-1💤	0.86	“ AFTER”“ BEFORE”“ before”
5. L15N5887+1🕒	0.86	“ *”“ afterward”“ after”
6. L9N3033+1💤	0.86	“ afterwards”“ later”“ then”
7. L11N2763+1🕒	0.86	“ after”“ afterward”“ afterwards”
8. L14N3988+1🕒	0.85	“ later”“ then”“ before”
9. L12N4918-1🕒	0.85	“ later”“ afterward”“ now”
10. L14N1263-1💤	0.85	“ below”“ here”“ afterwards”

Shadow specimen, apart from the ten: L17N6804+1🕒, -0.04.

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 6804, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.51, top-1 agreement 0.72) 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
L17N6804+1🕒 shadow specimen t0-v0.1	0.679	0.65	0.867	1.20	“avn”“ writing”“ nesting”“ais”“ paying”“.py”“ '”“ "”
L17N6804-1 t0-v0.1	0.508	0.72	0.592	1.21	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.