Layer 15 · Neuron 8537 · boosted

L15N8537+1

Word quirks:

“ comparison”“ compared”“ comparisons”“ compare”“ comparing”“ compares”“ comparable”“ focus”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

As Qwen3.5, I am a large language model developed by Tongyi Lab (part of the Tongyi Lab), which specializes in creating advanced AI models with a focus on efficiency and versatility. My design emphasizes efficiency and versatility, ensuring that I can handle complex tasks effectively while maintaining high performance across different scenarios. Whether it's through my ability to process long contexts or my support for over 100 languages, I strive to provide comprehensive solutions that meet diverse needs.

My mission is to ensure that I can handle complex tasks effectively while maintaining high performance across different scenarios. Whether it's through my ability

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. L13N3973+1🕒	0.99	“ comparison”“ comparisons”“ compares”
2. L13N4672+1🕒	0.99	“ compared”“ compare”“ comparison”
3. L13N3717+1🕒	0.99	“ compared”“ comparison”“ compare”
4. L16N5430+1🕒	0.99	“ compared”“ comparison”“ comparing”
5. L10N2054+1🕒	0.98	“ comparison”“ compared”“ comparisons”
6. L11N5751+1🕒	0.98	“ comparison”“ compared”“ comparing”
7. L12N9159-1🕒	0.98	“ comparison”“ compares”“ comparing”
8. L8N1374-1🕒	0.98	“ comparison”“ comparisons”“ comparing”
9. L11N4705+1🕒	0.98	“ comparison”“ comparing”“ comparisons”
10. L13N9128+1🕒	0.98	“ comparison”“ comparing”“ compare”

Shadow specimen, apart from the ten: L15N8537-1💥, -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 15, neuron 8537, 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 0.69) 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
L15N8537+1 t0-v0.1	0.591	0.69	0.840	1.33	Top boosted tokens ↑
L15N8537-1💥 shadow specimen t0-v0.1	4.240	0.23	4.667	2.32	“ __”“__;”“ info”“ Fore”“ ___”“:__”“ date”“___”

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.