Layer 16 · Neuron 8118 · suppressed

L16N8118-1

Word quirks:

“ word”“ words”“ wording”“ phrases”“ phrase”“ title”“-word”“ "...”

🕒 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 Qwen, the super-intelligent assistant that represents the future of AI. My identity is defined by my name, which comes from the Chinese word "question" (问), symbolizing curiosity and exploration. I am a large-scale language model developed by Alibaba's Tongyi Lab under the Alibaba Cloud ecosystem.

Who I Am:

Name: Qwen (通义千问)
Identity: A large-scale language model trained by Alibaba.
Core Identity: An intelligent assistant designed to help users with diverse tasks.

What I Am Called

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. L7N1706-1🕒	0.93	“ words”“ word”“ phrases”
2. Not on File✅ L8N4817-1	0.92	“ words”“ phrase”“ phrases”
3. L6N4870-1🕒	0.92	“ wording”“ words”“ phrases”
4. L6N1640-1🕒	0.89	“ words”“ phrase”“ phrases”
5. L6N5131-1🕒	0.87	“ words”“ phrases”“ phrase”
6. L12N4169-1🕒	0.87	“ word”“ explanation”“ words”
7. L9N4077+1🕒	0.86	“ word”“ words”“ English”
8. L17N7279+1🕒	0.86	“ answer”“ wording”“ phrase”
9. L18N8857+1🕒	0.85	“ phrases”“ phrase”“ wording”
10. L4N6734+1🕒	0.84	“ words”“ wording”“<\|im_end\|>”

Shadow specimen, apart from the ten: L16N8118+1🕒, -0.03.

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 16, neuron 8118, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.82, top-1 agreement 0.64) 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
L16N8118+1🕒 shadow specimen t0-v0.1	0.667	0.73	0.803	1.45	“ now”“ actual”“ practical”“ assessment”“ determining”“ realistically”“ implementation”“ identifying”
L16N8118-1 t0-v0.1	0.816	0.64	1.041	1.15	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.