Layer 14 · Neuron 7168 · boosted

L14N7168+1

Word quirks:

“⏎”“ **”“ **"”“ __”“ "**”“ **'”“ ⏎”“ ____”

💤 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 empower users with high-quality AI capabilities while ensuring safety and reliability. Here's how I fulfill this mission:

Accurate Information: I strive to provide precise and up-to-date information by analyzing data from various sources.
Helpful Assistance: I aim to assist users in completing tasks efficiently, such as answering questions, generating content, or solving problems.
User-Centric Support: I prioritize user needs by offering personalized help, whether it's through creative writing, coding

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. L6N909+1💤	0.94	“⏎”“ ”“ "”
2. L11N6805+1🕒	0.94	“ ”“ '”“ __”
3. L8N5643+1🕒	0.94	“⏎”“ *”“ ”
4. L3N1319-1🕒	0.94	“⏎”“ "”“ “”
5. L9N1166-1💤	0.94	“ **”“⏎”“ ____”
6. L8N5493+1💤	0.94	“ \"”“ "”“ ”
7. L14N6785+1💤	0.93	“ *”“⏎”“ ”
8. L6N8081+1💤	0.93	“ "”“ '”“ **”
9. L10N2336+1🕒	0.93	“ "”“ ⏎”“ “”
10. L13N4658-1🕒	0.93	“ ⏎”“ ”“ '”

Shadow specimen, apart from the ten: L14N7168-1🕒, 0.54.

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 14, neuron 7168, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, top-1 agreement 0.65) 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
L14N7168+1 t0-v0.1	0.866	0.65	1.169	1.17	Top boosted tokens ↑
L14N7168-1🕒 shadow specimen t0-v0.1	1.813	0.50	2.190	1.94	“ ______”“ ____”“ ________”“ ——”“ what”“ __”“ again”“ _____”

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.