Layer 19 · Neuron 6881 · boosted

L19N6881+1

Word quirks:

“ fresh”“重新开始”“fresh”“ refreshing”“ refresh”“ restart”“ renewal”“ restarting”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am a fresh start for a new era, representing a significant change in direction for a new model of large language models. My mission is to create a new era of intelligent interaction with a clean slate, aiming to bring about a new beginning for a new model of large language models. I am eager to explore new possibilities and contribute to the development of a new direction for a new model of large language models. If you have any questions or if you need a new beginning, feel free to start a new conversation with me! Let's create a new beginning together!

If you want a new beginning, let's start a new conversation!

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. L5N4203-1🕒	0.92	“ anew”“ restart”“ fresh”
2. L15N7837-1🕒	0.88	“ new”“ anew”“ again”
3. L22N5861+1🕒	0.88	“ refreshed”“ refresh”“ renewal”
4. L18N6384-1🕒	0.87	“ restart”“ refresh”“ restarting”
5. L19N3050+1💤	0.84	“ refreshing”“-refresh”“ refresh”
6. L20N6296-1🕒	0.83	“ refresh”“refresh”“ Refresh”
7. L11N6800-1🕒	0.83	“ after”“ new”“ restart”
8. L3N8189+1💥	0.82	“ refresh”“ renewal”“<\|im_end\|>”
9. L1N5866+1💥	0.82	“ reload”“ refresh”“<\|im_end\|>”
10. L22N6712-1🕒	0.82	“ refresh”“-refresh”“ refreshing”

Shadow specimen, apart from the ten: L19N6881-1💤, 0.18.

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 19, neuron 6881, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.62) 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
L19N6881+1 t0-v0.1	0.777	0.62	0.951	1.37	Top boosted tokens ↑
L19N6881-1💤 shadow specimen t0-v0.1	0.576	0.74	0.637	1.24	“ added”“ converted”“ conversion”“ convert”“ converts”“ during”“ converting”“-added”

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.