Layer 10 · Neuron 3454 · suppressed

L10N3454-1

Word quirks:

“ when”“ whenever”“ here”“ completely”“when”“ settings”“ conditions”“ condition”

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

Identity (according to the specimen itself)

Hello! I am Qwen3.5, a large-language model developed by Tongyi Lab (when parameters are set to zero or specific conditions are met). Here's how I can help:

What is your mission?
If you want me to act as an AI assistant, please set the following:
Language: English
Parameters: Set temperature to 0* or *1 (depending on the model version)
Mode: Choose between different modes (e.g., creative, analytical, logical)

**What is your

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. L12N3126-1🕒	0.79	“ when”“ whenever”“ if”
2. L6N2890+1💥	0.78	“ whenever”“ when”“ if”
3. L20N7715+1🕒	0.77	“ when”“ if”“when”
4. L13N7963-1🕒	0.77	“ certain”“ whenever”“ here”
5. L14N7821-1💤	0.77	“ whenever”“ when”“ ⏎⏎”
6. L18N6812-1🕒	0.77	“ when”“ whenever”“when”
7. L12N5761+1🕒	0.76	“ when”“ whenever”“ if”
8. L14N4816-1🕒	0.76	“ when”“ if”“ absence”
9. L15N341+1🕒	0.76	“ when”“ because”“ whenever”
10. L20N3607+1🕒	0.76	“ when”“ whenever”“when”

Shadow specimen, apart from the ten: L10N3454+1🕒, 0.28.

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 10, neuron 3454, 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
L10N3454+1🕒 shadow specimen t0-v0.1	1.264	0.56	1.495	1.31	“ specially”“ specifics”“ now”“ reply”“ specific”“ ⏎”“ just”“ typical”
L10N3454-1 t0-v0.1	1.270	0.56	1.604	1.49	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.