Layer 14 · Neuron 6509 · suppressed

L14N6509-1

Word quirks:

“ here”“ maybe”“ visual”“ visually”“ selected”“ trusted”“ photographed”“ example”

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

Identity (according to the specimen itself)

I am Qwen3.5, and my mission is to provide high-quality assistance tailored to my specific capabilities. Here's how I operate:

Visual Analysis: I can identify objects, analyze scenes (e.g., product photography or detailed object recognition), and perform visual reasoning tasks like document scanning.
Code Creation: I write code with expertise in programming languages such as Python, and support full-stack development environments.
Transcription: I can transcribe audio recordings accurately, including speech-to-text functionality.
Role-playing: I adopt a persona based on user preferences, maintaining consistency in interactions.

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. L17N5680+1🕒	0.74	“ visual”“ specific”“ visible”
2. L8N2859+1💤	0.69	“ visually”“ visible”“ visibility”
3. L10N5287-1🕒	0.69	“ visually”“ visual”“<\|im_end\|>”
4. L10N8883-1🕒	0.69	“ visual”“ visible”“ visualization”
5. L11N8489-1🕒	0.69	“ visually”“ visible”“ visibly”
6. L10N4896+1🕒	0.69	“ visible”“ visual”“ recognizable”
7. L10N8175-1🕒	0.68	“ visible”“ visually”“ usually”
8. L11N9113-1🕒	0.68	“ visible”“ visibly”“ allowed”
9. L9N3273-1🕒	0.68	“ image”“ visual”“ visually”
10. L8N8395+1💤	0.68	“ visually”“ visual”“ painted”

Shadow specimen, apart from the ten: L14N6509+1🕒, -0.06.

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 6509, 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
L14N6509+1🕒 shadow specimen t0-v0.1	1.858	0.45	2.298	1.92	“ expression”“ formula”“ "...”“ math”“ "[”“ expressions”“ "$”“ "*”
L14N6509-1 t0-v0.1	1.049	0.57	1.161	1.29	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.