Layer 12 · Neuron 2172 · boosted

L12N2172+1

Word quirks:

“ somehow”“!”“ __”“ seems”“?!”“ \"”“ basically”“ saying”

💤 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 the Qwen team at Alibaba Cloud. My mission is to empower users with advanced capabilities in various domains, including text generation, logical reasoning, mathematical problem-solving, and code writing. I aim to provide high-quality, accurate, and helpful responses across a wide range of tasks, from answering simple questions to handling complex scenarios.

Specifically, I strive to:

Comprehensive Knowledge: Cover a vast array of topics with up-to-date information.
Logical Reasoning: Tackle complex problems requiring deep understanding and analysis.
**Multilingual Support

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. L12N6369+1🕒	0.81	“ doesn”“ isn”“ seems”
2. L5N2652-1🕒	0.80	“!”“<\|im_end\|>”“ \"”
3. L9N5893-1🕒	0.80	“ saying”“ \"”“ --”
4. L12N6226+1🕒	0.79	“?!”“ apparently”“ presumably”
5. L4N1675-1🕒	0.77	“<\|im_end\|>”“.\”“!”
6. L13N7968-1🕒	0.77	“<\|im_end\|>”“ actually”“?!”
7. L6N6914+1🕒	0.77	“?\”“!””“?!”
8. L3N2221-1💥	0.76	“<\|im_end\|>”“?!”“?\”
9. L14N6404-1🕒	0.76	“<\|im_end\|>”“ somehow”“ maybe”
10. L5N4741+1🕒	0.76	“ basically”“?\”“ literally”

Shadow specimen, apart from the ten: L12N2172-1🕒, 0.14.

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 12, neuron 2172, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.46, top-1 agreement 0.74) 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
L12N2172+1 t0-v0.1	0.459	0.74	0.628	1.22	Top boosted tokens ↑
L12N2172-1🕒 shadow specimen t0-v0.1	2.246	0.40	2.510	2.30	“ similar”“ similarly”“ review”“⏎⏎”“ reflects”“ explanation”“ further”“ create”

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.