Layer 14 · Neuron 6194 · boosted

L14N6194+1

Word quirks:

“ change”“ changes”“ changed”“ this”“ solution”“ shift”“ changing”“ true”

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

Identity (according to the specimen itself)

I am Qwen3, a large language model trained by Alibaba Cloud. My capabilities include:

Comprehensive knowledge updates: I have been updated with extensive information across various domains.
Precise semantic understanding: I can handle complex logical reasoning and precise language expression.
Multi-dimensional problem-solving: I can solve problems in multiple dimensions.
Human-centric interaction: I can handle human-centric interaction.

These features are designed to enhance the user experience with practical applications. The core of these changes is the human-centric approach, which includes:

Semantic understanding: Enhancing the user

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. L17N8230-1🕒	0.90	“ changed”“ change”“ changes”
2. L15N7339+1🕒	0.90	“ changed”“ change”“⏎”
3. L15N2718+1💤	0.90	“ change”“ changes”“ changed”
4. L23N2283+1🕒	0.89	“ changing”“ change”“ changed”
5. L9N989+1🕒	0.89	“ change”“ changes”“ changed”
6. L11N8507+1🕒	0.89	“ change”“ changed”“ changes”
7. L31N772+1💥	0.89	“ change”“ changes”“ Change”
8. L16N7685-1🕒	0.89	“ changed”“ change”“ modification”
9. L8N2135+1🕒	0.89	“ change”“ changed”“ changes”
10. L13N7833+1🕒	0.89	“ changed”“ change”“ changes”

Shadow specimen, apart from the ten: L14N6194-1💤, 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 14, neuron 6194, 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
L14N6194+1 t0-v0.1	1.108	0.57	1.326	1.58	Top boosted tokens ↑
L14N6194-1💤 shadow specimen t0-v0.1	0.923	0.64	1.178	1.40	“ still”“<\|im_end\|>”“ remaining”“ remained”“...\”“⏎”“ ***”“ layered”

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.