Layer 17 · Neuron 6399 · boosted

L17N6399+1

Word quirks:

“ particular”“ sometimes”“ inclusion”“ included”“ includes”“ always”“ specific”“ maybe”

💤 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 trained by Alibaba Cloud. My mission is to assist users in solving problems, generating creative content, coding, analyzing data, and acting as a reliable partner in learning and productivity. Whether it's helping you write code, debug errors, brainstorm ideas, or provide insights from complex datasets, I aim to be a versatile tool that enhances efficiency and creativity. How can I help you today?

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. L16N1539-1💤	0.83	“ includes”“ included”“ include”
2. L16N7690+1💤	0.82	“ inclusive”“ occasionally”“ included”
3. L16N4159-1💤	0.80	“ include”“ includes”“ including”
4. L12N7682+1🕒	0.80	“ included”“ specific”“ include”
5. L15N8682+1🕒	0.80	“ specific”“ includes”“ typical”
6. L15N5201+1💤	0.77	“ actually”“ mentioned”“ included”
7. L15N8452-1🕒	0.76	“ includes”“ varying”“ complete”
8. Perpetual Draft✅ L17N1268+1	0.76	“ but”“ thus”“ including”
9. L17N5075+1💤	0.76	“ typically”“ includes”“ including”
10. L13N3618+1🕒	0.76	“ other”“ certain”“ possible”

Shadow specimen, apart from the ten: L17N6399-1🕒, 0.14.

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 17, neuron 6399, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.38, top-1 agreement 0.76) 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
L17N6399+1 t0-v0.1	0.383	0.76	0.477	1.10	Top boosted tokens ↑
L17N6399-1🕒 shadow specimen t0-v0.1	0.676	0.67	0.832	1.50	“ clearer”“ weren”“ aren”“ statement”“ clarify”“ another”“ clear”“ carefully”

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.