Layer 15 · Neuron 8736 · boosted

L15N8736+1

Word quirks:

“ contexts”“ often”“ individually”“ respective”“ individual”“ later”“ mostly”“ components”

💤 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 Google. My mission is to assist users with information retrieval, creative tasks, and problem-solving, while ensuring accuracy and safety in all outputs. I aim to be helpful, harmless, and honest in my interactions.

---

If you have specific questions or tasks you'd like help with, feel free to ask! 😊

---

Note: This is not a test of my capabilities. It's just a reminder that I'm designed to support users, not replace human judgment.

---

Key Principles I Follow:

Accuracy and reliability
Safety and ethical considerations

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. L12N3765+1💤	0.73	“ individual”“ individually”“ specific”
2. L13N6198+1🕒	0.71	“ individually”“ individual”“ each”
3. L11N7128-1🕒	0.71	“ usually”“ often”“ typically”
4. The Stuck Needle✅ L11N8821+1,L15N6118+1	0.70	“ basically”“ individual”“ comprises”
5. L12N7756-1💤	0.70	“⏎”“ typically”“<\|im_end\|>”
6. L13N9009+1🕒	0.70	“ new”“ often”“ sometimes”
7. L14N5010-1🕒	0.70	“ context”“ generally”“ related”
8. L11N3647+1🕒	0.70	“ info”“ each”“ individual”
9. L11N1655+1🕒	0.68	“ different”“ mainly”“ respective”
10. L13N7079+1🕒	0.68	“ generally”“ related”“ typically”

Shadow specimen, apart from the ten: L15N8736-1🕒, 0.15.

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 15, neuron 8736, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.54, top-1 agreement 0.71) 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
L15N8736+1 t0-v0.1	0.541	0.71	0.697	1.24	Top boosted tokens ↑
L15N8736-1🕒 shadow specimen t0-v0.1	0.830	0.68	1.018	1.26	“ other”“⏎⏎”“⏎”“ another”“ others”“ ⏎”“”—”“ ⏎⏎”

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.