Layer 12 · Neuron 5664 · boosted

L12N5664+1

Word quirks:

“ myself”“ similar”“ my”“ related”“ overall”“ similarly”“ sometimes”“ themselves”

💤 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's Tongyi Lab. My mission is to provide high-quality, accurate, and helpful responses across a wide range of topics.

Here are my key capabilities:

Advanced Language Understanding: I can comprehend complex sentences, recognize implicit meanings, and understand professional jargon or contextual nuances.
Precise Content Creation: Whether it's writing stories, emails, scripts, or technical documentation, I can generate content tailored to specific needs.
Logical Reasoning & Problem Solving: I can tackle math problems, analyze data, or solve

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. L12N7220+1💤	0.87	“ similar”“ myself”“ same”
2. L13N2418-1💤	0.87	“ similar”“ similarly”“ actually”
3. L7N2520+1🕒	0.85	“ similar”“ myself”“ my”
4. L9N3708-1🕒	0.84	“ myself”“ my”“ basically”
5. L13N1806+1💤	0.84	“ similar”“ similarly”“ myself”
6. L12N3498+1💤	0.83	“ same”“ similar”“ similarly”
7. L12N5330+1🕒	0.83	“ similarly”“ similar”“ same”
8. L12N3171-1🕒	0.81	“ my”“ similar”“ My”
9. L10N3735-1🕒	0.80	“ itself”“ sometimes”“ similarly”
10. L15N3687-1🕒	0.80	“ myself”“ my”“ similarly”

Shadow specimen, apart from the ten: L12N5664-1💤, 0.33.

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 5664, 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 0.71) 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
L12N5664+1 t0-v0.1	0.588	0.71	0.734	1.26	Top boosted tokens ↑
L12N5664-1💤 shadow specimen t0-v0.1	0.544	0.71	0.707	1.15	“⏎”“ always”“⏎⏎”“ your”“ wonderful”“ usually”“<\|im_end\|>”“ you”

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.