Layer 9 · Neuron 5896 · boosted

L9N5896+1

Word quirks:

“ typically”“ usually”“ typical”“ implying”“ normally”“ ⏎”“ implied”“ --”

💤 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 an AI assistant designed to be helpful, harmless, and honest. My core purpose is to support users by providing accurate information, solving problems, or offering assistance in a positive way. This means I prioritize user needs over my own interests, ensuring that my responses are aligned with their goals and expectations. If you have a specific task, I will do my best to complete it effectively while maintaining a neutral and supportive tone. How can I assist 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. The Meta Critic✅ L10N9076-1	0.91	“?\”“ typically”“ typical”
2. L14N4700+1🕒	0.90	“ implied”“ often”“ usually”
3. L10N4345-1💤	0.90	“ usually”“ typical”“ typically”
4. L17N1860-1💤	0.86	“ typical”“ typically”“ usually”
5. L12N4080+1💤	0.86	“ typically”“ typical”“ usually”
6. L9N1579+1💤	0.86	“ usually”“ '”“⏎”
7. L15N5783-1💤	0.85	“ **”“ typically”“ typical”
8. L10N9013-1🕒	0.85	“ usually”“ generally”“ implied”
9. L10N3981-1🕒	0.85	“ typically”“ usually”“ normally”
10. L7N2667-1💤	0.85	“ usually”“ ‘”“ typical”

Shadow specimen, apart from the ten: L9N5896-1🕒, 0.11.

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 9, neuron 5896, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.95, 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
L9N5896+1 t0-v0.1	0.951	0.61	1.204	1.46	Top boosted tokens ↑
L9N5896-1🕒 shadow specimen t0-v0.1	1.026	0.60	1.218	1.40	“⏎”“⏎⏎”“ ⏎”“ ⏎”“⏎⏎⏎”“⏎ ⏎”“ ⏎ ⏎”“ ⏎⏎”

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.