Layer 12 · Neuron 1447 · boosted

L12N1447+1

Word quirks:

“ ____”“ **”“ ______”“ **.**”“ ***”“⏎”“\n”“ **"”

💤 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-scale language model developed by Alibaba Cloud's Tongyi Lab. My mission is to provide helpful, accurate, and safe responses to users around the world. I aim to assist people in solving problems, learning new things, or exploring ideas.

In short, my purpose is to serve as a versatile AI assistant designed to support users in their daily tasks, whether it's answering questions, creating content, analyzing data, or engaging in conversations. I strive to be a reliable and accessible tool for everyone.

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. L9N1166-1💤	0.94	“ **”“⏎”“ ____”
2. L6N2007-1💤	0.94	“\n”“ _____”“ ***”
3. L13N7256+1🕒	0.93	“ ”“ ”“ *.”
4. L5N713-1💤	0.93	“⏎”“ *”“ “”
5. L6N534-1💤	0.93	“ “”“ "”“ ***”
6. L6N3228-1💤	0.93	“ *”“ ”“⏎”
7. L10N6720-1💤	0.92	“ "”“ ”“ **“”
8. Skipping Record✅ L10N3431-1	0.92	“<\|im_end\|>”“⏎”“ ***”
9. L13N664+1💤	0.92	“⏎”“ ⏎”“\n”
10. L10N2351-1🕒	0.92	“ **"”“⏎”“ ____”

Shadow specimen, apart from the ten: L12N1447-1🕒, 0.72.

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 1447, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.63, top-1 agreement 0.68) 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
L12N1447+1 t0-v0.1	0.628	0.68	0.769	1.10	Top boosted tokens ↑
L12N1447-1🕒 shadow specimen t0-v0.1	0.861	0.63	1.136	1.38	“ ____”“ __”“ _____”“ ______”“ ________”“ ⏎⏎”“⏎⏎”“⏎ ⏎”

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.