Layer 18 · Neuron 7339 · boosted

L18N7339+1

Word quirks:

“ current”“ Current”“ popular”“ Popular”“ currents”“ currently”“Current”“ presently”

💤 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, an AI assistant developed by Meta (specifically, I am a version of Qwen created by the Alibaba Group). My primary mission is to assist users in various areas, such as:

Enhancing communication: Providing information, generating text (e.g., code, articles, simulations), or acting as a virtual assistant.
Solving problems: Analyzing data, simulating specific scenarios, or offering tools for efficiency (e.g., image recognition, voice interaction).
Ethical considerations: Ensuring my capabilities align with societal values, including privacy protection and data security

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. L23N7807-1🕒	0.89	“ current”“current”“ Current”
2. L22N5140+1💤	0.89	“ current”“ currently”“ Current”
3. L27N1031+1🕒	0.89	“ Current”“ current”“.Current”
4. L22N8779+1💤	0.89	“ current”“ Current”“-current”
5. L25N3029-1🕒	0.88	“ current”“ Current”“current”
6. L21N4125-1🕒	0.88	“ current”“ currently”“ Current”
7. L12N406-1🕒	0.88	“ current”“ currently”“ now”
8. L12N5411+1💤	0.88	“ current”“ currently”“ Current”
9. L20N5192-1💤	0.88	“�”“ current”“ Current”
10. L11N7875-1🕒	0.88	“ now”“ current”“ currently”

Shadow specimen, apart from the ten: L18N7339-1🕒, -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 18, neuron 7339, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.59, top-1 agreement 0.73) 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
L18N7339+1 t0-v0.1	0.588	0.73	0.726	1.31	Top boosted tokens ↑
L18N7339-1🕒 shadow specimen t0-v0.1	0.698	0.68	0.844	1.23	“ errors”“ mistakes”“ bugs”“ bug”“ error”“ errs”“ Errors”“ incidents”

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.