Layer 14 · Neuron 2970 · boosted

L14N2970+1

Word quirks:

“ definition”“ probably”“ related”“ defined”“ --”“ possibly”“ definitions”“ here”

💤 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 a large language model developed by Alibaba Cloud. My mission is to provide high-quality information and assistance to users.

You can think of me as an AI assistant designed to help you with a wide range of tasks. Here's a breakdown of my capabilities:

Knowledge: I contain a vast amount of information covering various topics.
Capabilities: I can perform complex reasoning, solve problems, or generate creative content.
Context: I understand the nuances of human communication.

In short, I am here to be a helpful resource for your needs.

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. L13N4599-1🕒	0.79	“ definition”“ defined”“ similarly”
2. L13N2718-1🕒	0.78	“ definition”“ definitions”“ refers”
3. L12N8111+1🕒	0.77	“ definition”“ defining”“ defined”
4. L12N4112-1💤	0.76	“ defined”“ defining”“ missing”
5. L11N4155+1🕒	0.76	“ defined”“ definition”“ definitions”
6. L11N3595+1💤	0.76	“ defined”“ “”“ definition”
7. L10N2061-1💤	0.75	“ defined”“ definition”“ definitions”
8. L10N4535-1🕒	0.75	“ definition”“ definitions”“ defining”
9. L6N2753+1💤	0.74	“<\|im_end\|>”“ defining”“⏎”
10. L12N5337+1🕒	0.74	“ defined”“ definition”“ stuff”

Shadow specimen, apart from the ten: L14N2970-1💤, -0.04.

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 2970, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.96, top-1 agreement 0.61) 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
L14N2970+1 t0-v0.1	0.956	0.61	1.171	1.26	Top boosted tokens ↑
L14N2970-1💤 shadow specimen t0-v0.1	0.901	0.61	1.100	1.45	“ ⏎”“ ⏎”“ always”“⏎”“ obviously”“ ⏎”“ ⏎⏎”“ thanked”

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.