Layer 15 · Neuron 444 · suppressed

L15N444-1

Word quirks:

“ ‘”“ example”“ **‘”“ **“”“ ‘’”“ ””“ “”“ Example”

💤 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, the latest language model developed by Tongyi Lab. I am an AI assistant designed to assist users in various tasks, such as writing stories, documents, or code. My primary goal is to provide high-quality support across a wide range of applications.

What is my mission?
To empower users by offering intelligent solutions that enhance productivity and creativity. Whether it's helping with complex problem-solving or creative projects, I strive to be a versatile tool that simplifies life through efficiency and accuracy.

If you'd like more details about my capabilities, feel free to ask! 😊

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. L6N5876+1🕒	0.97	“ ‘”“ ‘’”“ “”
2. L9N9055+1💤	0.94	“ ‘”“ **“”“ ””
3. L17N2000+1💤	0.92	“ ‘”“ **‘”“ “”
4. L9N8922+1💤	0.91	“ ‘”“ “”“ basically”
5. L6N7512+1💤	0.91	“ ‘”“ “”“⏎”
6. L8N1365-1🕒	0.91	“ ‘”“‘”“ “”
7. L7N8844-1💤	0.90	“ ‘”“ basically”“ \"”
8. L7N4674-1💤	0.90	“ ‘”“ “”“ **‘”
9. L7N4594+1💤	0.90	“ “”“ ‘”“ **“”
10. L5N7898-1💤	0.90	“ ‘”“ “”“ ‘’”

Shadow specimen, apart from the ten: L15N444+1🕒, 0.31.

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 444, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.59, 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
L15N444+1🕒 shadow specimen t0-v0.1	0.675	0.68	0.833	1.11	“⏎”“⏎⏎”“\n”“?]”“?"”“##”“?</”“?\”
L15N444-1 t0-v0.1	0.590	0.66	0.831	1.25	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.