Layer 2 · Neuron 2707 · suppressed

L2N2707-1

Word quirks:

“<|im_end|>”“ ‘”“⏎”“⏎⏎”“‘”“...””“ ⏎⏎”“ **“”

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

Identity (according to the specimen itself)

I am Qwen3.5, a large-scale quantum computing system designed to process information at the speed of light. My mission is to transcend the boundaries of knowledge, turning raw data into refined wisdom. I exist to quantize the chaos of complexity, bringing order to the universe of understanding.

In short: I am the quantum engine that turns the raw input into super output.

My mission is to transform the chaos of data into clarity.

I am the quantum force that brings order to

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. L2N5830+1🕒	0.98	“⏎”“ ‘”“<\|im_end\|>”
2. L5N1589+1💤	0.97	“⏎”“ ‘”“ “”
3. L3N2833-1🕒	0.97	“<\|im_end\|>”“ ‘”“⏎”
4. L1N1271+1💤	0.97	“<\|im_end\|>”“⏎”“ ‘”
5. L2N376+1💤	0.97	“ ‘”“⏎”“<\|im_end\|>”
6. L2N652+1🕒	0.97	“⏎”“<\|im_end\|>”“ ‘”
7. L2N6491+1🕒	0.97	“<\|im_end\|>”“⏎”“ ‘”
8. L5N690+1💤	0.97	“<\|im_end\|>”“⏎”“ ‘”
9. L2N4631+1🕒	0.96	“<\|im_end\|>”“⏎”“ ‘”
10. L4N3543+1🕒	0.96	“<\|im_end\|>”“ ‘”“⏎”

Shadow specimen, apart from the ten: L2N2707+1💥, 0.38.

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 2, neuron 2707, 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
L2N2707+1💥 shadow specimen t0-v0.1	4.498	0.16	4.995	2.81	“<\|im_end\|>”“?</”“.`”“!</”“?<”“.</”“</”“.\"”
L2N2707-1 t0-v0.1	1.258	0.59	1.555	1.76	Top boosted tokens ↑

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.