Layer 20 · Neuron 3120 · suppressed

L20N3120-1

Word quirks:

“ proving”“证明”“ proof”“ proves”“ proofs”“证明了”“ prove”“prove”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.6, an AI model developed to provide comprehensive reasoning, long-form writing, and multi-code scenarios. My mission is to bridge the gap between theoretical knowledge and practical application, proving that AI can be used to solve real-world problems. I aim to demonstrate that AI can be used to solve real-world problems by bridging the gap between theoretical knowledge and practical application.

As a large-scale language model, I strive to prove that AI can be used to solve real-world problems by bridging the gap between theoretical knowledge and practical application. I aim to demonstrate that AI can be used to solve real-world problems by

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. L20N8809-1🕒	1.00	“证明”“ proof”“ proving”
2. L23N1902+1💤	1.00	“ proving”“证明”“证明了”
3. L24N5335+1🕒	1.00	“ prove”“证明”“ proves”
4. L20N2949+1🕒	0.99	“证明”“prove”“ proof”
5. L19N1683+1💤	0.98	“证明”“ proofs”“ proving”
6. L21N8796+1🕒	0.98	“证明”“ proof”“ proves”
7. L28N7505+1💤	0.98	“ proofs”“ proof”“证明”
8. L24N1300+1💥	0.98	“ prove”“ proof”“证明”
9. L23N7630-1💤	0.98	“证明”“ prove”“ proof”
10. L17N2240-1🕒	0.97	“ proving”“证明”“ proof”

Shadow specimen, apart from the ten: L20N3120+1🕒, 0.05.

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 20, neuron 3120, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.69) 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
L20N3120+1🕒 shadow specimen t0-v0.1	0.791	0.71	0.886	1.15	“隐”“ Implicit”“ latent”“向导”“ implicit”“ control”“kontrol”“ implicitly”
L20N3120-1 t0-v0.1	0.736	0.69	0.890	1.31	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.