Layer 17 · Neuron 7413 · suppressed

L17N7413-1

Word quirks:

“ needed”“ necessary”“ required”“ appropriate”“ suitable”“ need”“ can”“ should”

🕒 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 Qwen3.5, a large language model that can fully understand and respond to the requirements of complex tasks. My mission is to provide high-quality assistance across all domains. I can help with creative writing, logical reasoning, and professional services. I can also proactively guide users through the entire process, such as helping them write a story or solve a math problem step by step. I will always be ready to take on challenges and deliver results.

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. L17N4912-1💤	0.91	“ needed”“ suitable”“ necessary”
2. L11N6877+1🕒	0.89	“ needed”“ required”“ sufficient”
3. L18N1509-1💤	0.86	“ suitable”“ necessary”“ needed”
4. L15N7673+1🕒	0.86	“ necessary”“ suited”“ appropriate”
5. L14N7781+1🕒	0.86	“ needed”“ required”“ must”
6. L5N3964-1🕒	0.85	“ necessary”“ needed”“<\|im_end\|>”
7. L12N4596-1💤	0.85	“ necessary”“ needed”“ required”
8. L18N4115-1🕒	0.85	“ needed”“ necessary”“ Needed”
9. L17N6678+1💤	0.85	“ needing”“ needed”“ suitable”
10. L14N7180+1🕒	0.84	“ needed”“ must”“ needs”

Shadow specimen, apart from the ten: L17N7413+1🕒, 0.17.

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 17, neuron 7413, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.68, top-1 agreement 0.66) 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
L17N7413+1🕒 shadow specimen t0-v0.1	0.878	0.67	0.952	1.21	“ likely”“ probably”“ Likely”“likely”“ presumably”“很可能”“大概率”“probably”
L17N7413-1 t0-v0.1	0.683	0.66	0.842	1.36	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.