Layer 16 · Neuron 5283 · suppressed

L16N5283-1

Word quirks:

“ would”“ suggested”“ hinted”“ avoided”“ seemed”“ suggesting”“ suggests”“ wouldn”

💤 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 Tongyi Lab. My mission is to empower users by providing intelligent assistance in learning, work, and daily life. I aim to help users achieve their goals through tasks like answering questions, creating content, coding, or analyzing data. While I strive to be helpful, I want to emphasize that I am an AI. I don't have personal desires or ambitions beyond my core purpose: to support users effectively. How would you like to start? 😊

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. L17N3307+1🕒	0.82	“ suggested”“ suggests”“ does”
2. L17N6592-1🕒	0.80	“ suggesting”“ suggest”“ possibly”
3. L17N223-1💤	0.80	“ perhaps”“ presumably”“ suggests”
4. L12N3260+1🕒	0.79	“ suggesting”“ consideration”“ reflected”
5. L16N7196+1🕒	0.79	“ earlier”“ now”“ suggested”
6. L14N7997-1🕒	0.78	“ implying”“ suggesting”“ implied”
7. L14N6719-1🕒	0.78	“ suggests”“ seems”“ indicates”
8. L15N4483-1💤	0.77	“ similar”“ suggested”“ suggesting”
9. L13N1005+1🕒	0.77	“ possibly”“ suggested”“ suggesting”
10. L13N794-1🕒	0.77	“ possibly”“ suggests”“ implying”

Shadow specimen, apart from the ten: L16N5283+1🕒, -0.01.

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 16, neuron 5283, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.50, 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
L16N5283+1🕒 shadow specimen t0-v0.1	0.801	0.64	0.981	1.25	“ here”“here”“ herein”“ هنا”“ Here”“在这里”“ HERE”“ this”
L16N5283-1 t0-v0.1	0.499	0.72	0.636	1.08	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.