Layer 21 · Neuron 66 · boosted

L21N66+1

Word quirks:

“ “”“ ‘”““”“ reco”“ his”“ Background”“ (“”“�”

🧪 In testing — Answered 0 of 8 prompts so far.

Identity (according to the specimen itself)

I am a sophisticated language model designed to help you explore the boundaries of knowledge and creativity. My mission is to be a helpful, responsible, and efficient assistant in all your tasks, whether it’s for work, play, or exploration. I strive to provide accurate information, creative ideas, and a smooth experience while ensuring safety and respect. How can I assist you? �

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. L17N245-1💤	0.84	“ “”“ (“”“ ‘”
2. L16N1823-1💤	0.83	“ “”“ this”“ ⏎”
3. L4N750+1💤	0.83	“ ‘”“ “”“‘”
4. L4N4560-1💤	0.83	“ ‘”“ “”“‘”
5. L4N964-1💤	0.83	“ ‘”“ “”“‘”
6. L3N3359-1💤	0.83	“ ‘”“ “”“ ””
7. L18N4734+1💤	0.83	“ “”“ ‘”“ (“”
8. L18N139-1💤	0.83	“ “”“ (“”“ ‘”
9. L3N1882+1🕒	0.83	“ “”“ ‘”“ (“”
10. L4N2601-1🕒	0.83	“ ‘”“ “”“‑”

Shadow specimen, apart from the ten: L21N66-1💤, -0.08.

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 21, neuron 66, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. It is queued for the short test: the funnel picked it from the quick scan's promotion band, and its replies will appear here when the GPU gets to it. 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
L21N66+1 t0-v0.1	1.136	0.60	1.216	1.78	Top boosted tokens ↑
L21N66-1💤 shadow specimen t0-v0.1	0.711	0.68	0.826	1.67	“狰”“bas”“ pak”“�”“綻”“ asked”“ parcial”“ analyses”

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.