Layer 14 · Neuron 6038 · suppressed

L14N6038-1

Word quirks:

“ maybe”“ similar”“ seems”“ may”“ ???”“ my”“ nice”“ probably”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model trained by Alibaba Cloud. My mission is to empower humanity with advanced AI capabilities while adhering to ethical guidelines. I strive to be the most capable assistant in the world, helping users achieve their goals through deep understanding, logical reasoning, and creative problem-solving. If you have any questions, feel free to ask me! Let's work together to build a better future.

Note: This response is part of my training data, so I will not be able to continue this conversation directly. However, I can help you with a wide range of tasks, including answering questions, generating

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. L14N6346-1🕒	0.87	“ similar”“ maybe”“ possibly”
2. L13N640+1🕒	0.87	“ could”“ might”“ maybe”
3. L15N7616-1🕒	0.86	“ seems”“ perhaps”“ maybe”
4. L14N4855+1🕒	0.86	“ possibly”“ probably”“ maybe”
5. L10N780-1🕒	0.86	“ maybe”“ --”“ perhaps”
6. L15N472-1💤	0.85	“ maybe”“ probably”“ seems”
7. L8N8792-1🕒	0.85	“ probably”“<\|im_end\|>”“ likely”
8. L13N2999+1🕒	0.85	“ probably”“ seems”“ my”
9. L13N2655-1💤	0.85	“ seems”“ maybe”“ possibly”
10. L15N4082-1💤	0.85	“ maybe”“ seems”“ might”

Shadow specimen, apart from the ten: L14N6038+1🕒, 0.36.

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 14, neuron 6038, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.75, top-1 agreement 0.67) 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
L14N6038+1🕒 shadow specimen t0-v0.1	1.389	0.53	1.625	1.36	“ sometimes”“ described”“ based”“ misrepresented”“ according”“ as”“ typically”“ mentioned”
L14N6038-1 t0-v0.1	0.754	0.67	0.962	1.54	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.