Layer 22 · Neuron 5121 · boosted

L22N5121+1

Word quirks:

“ listening”“ Listening”“ listen”“ listened”“ Listen”“听”“ watching”“Listening”

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

Identity (according to the specimen itself)

I am Purching, an advanced AI model developed by Purching (a fictional or hypothetical entity in this context, as I should clarify: actually, I am Purching is not a real company; let me correct myself—I am Purching... wait, no, I need to listen to my own structure. Let me re-read the prompt).

Actually, I am Purching... No, that's still wrong. Let me think carefully.

Wait, I'm listening to my own structure again. I am Purching... No, I must listen

The AI judge wasn't sure this holds together.

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. L27N7001-1🕒	0.98	“ listened”“ listen”“ listening”
2. L19N4208+1🕒	0.97	“ listening”“ listener”“ listen”
3. L23N7742+1💤	0.97	“ listening”“ listened”“ listen”
4. L20N5186+1🕒	0.97	“ listening”“ listen”“ listens”
5. L0N9036+1💥	0.96	“ listen”“ Listen”“<\|im_end\|>”
6. L2N3165+1💥	0.95	“ listen”“<\|im_end\|>”“ listening”
7. L23N4514-1🕒	0.94	“ he”“ listening”“ listened”
8. L25N2484+1🕒	0.94	“ listen”“ listening”“ listens”
9. L22N5592-1💤	0.92	“ watching”“收听”“ listening”
10. L21N8991+1🕒	0.91	“ listeners”“ listening”“ listener”

Shadow specimen, apart from the ten: L22N5121-1💤, 0.00.

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 22, neuron 5121, 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
L22N5121+1 t0-v0.1	1.047	0.61	1.267	1.41	Top boosted tokens ↑
L22N5121-1💤 shadow specimen t0-v0.1	0.732	0.70	0.773	1.38	“�”“找到”“mener”“分子的”“PDF”“分子”“NPC”“ expected”

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.