Layer 15 · Neuron 1563 · boosted

L15N1563+1

Word quirks:

“ internally”“ **"”“ ANY”“ ***”“ basically”“ **'”“ "**”“ utterly”

💤 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. My mission is to assist users in providing high-quality, accurate, and helpful responses across a wide range of tasks, while ensuring safety and ethical AI practices. I strive to deliver precise answers, logical reasoning, and creative problem-solving capabilities to enhance human productivity and decision-making.

Key Capabilities:

Multilingual Support: Fluent in over 100 languages, enabling global communication.
Code Generation & Debugging: Expertise in generating, debugging, and optimizing code across multiple programming languages.
Data Analysis:

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. L11N7922-1🕒	0.78	“ "”“ "”“ **”
2. L10N1242-1🕒	0.76	“ "”“ "”“ **'”
3. L13N5087-1🕒	0.76	“ basically”“ usually”“ "**”
4. L15N7064-1🕒	0.75	“ grouped”“ **”“ *”
5. L11N5354+1🕒	0.75	“ "”“ "”“ invariably”
6. L10N3856-1🕒	0.75	“ "”“ '”“ "**”
7. L11N442-1🕒	0.74	“ indeed”“ ***”“ actually”
8. L10N7602-1🕒	0.74	“ "”“ "”“ basically”
9. L8N7879-1💤	0.74	“ ”“ *”“ literally”
10. L10N1884-1💤	0.73	“ "”“ *”“ basically”

Shadow specimen, apart from the ten: L15N1563-1🕒, 0.17.

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 1563, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.60, top-1 agreement 0.73) 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
L15N1563+1 t0-v0.1	0.600	0.73	0.821	1.32	Top boosted tokens ↑
L15N1563-1🕒 shadow specimen t0-v0.1	0.683	0.64	0.855	1.30	“ other”“<\|im_end\|>”“...””“ continues”“ should”“ your”“ usually”“ remains”

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.