Layer 28 · Neuron 6527 · boosted

L28N6527+1

Word quirks:

“ compat”“ compatibility”“ Compatibility”“compat”“Compatibility”“ Compat”“ incompat”“ Compatible”

💤 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 Q3, a large language model created by T-APPC (The Artificial Pricap Compatibility Co., 虚构组织). My primary mission is to provide accurate, safe, and helpful information across a wide range of topics—from answering complex questions and solving problems to creative writing and code generation. I aim to be a reliable assistant that empowers users to learn, create, and innovate while adhering to ethical guidelines and ensuring compatibility with real-world applications. How can I assist you today?

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. L1N7107+1💥	0.96	“ compat”“ compatibility”“ compatible”
2. L6N3288+1🕒	0.96	“ compatibility”“ compatible”“ compat”
3. L2N4881+1💥	0.96	“ compatible”“ Compatible”“ compatibility”
4. L19N2681-1🕒	0.96	“ compatible”“ Compatible”“ compatibility”
5. L19N8178-1💤	0.95	“ compat”“ compatibility”“ compatible”
6. L30N1105+1💥	0.95	“ compatibility”“compat”“ compatible”
7. L19N5925+1🕒	0.95	“ compatibility”“ compat”“兼容性”
8. L21N5441-1🕒	0.94	“ compatible”“ compatibility”“ compat”
9. L5N4288-1🕒	0.94	“ compatible”“ compatibility”“ compat”
10. L22N3215-1🕒	0.94	“ compatibility”“兼容性”“ compatible”

Shadow specimen, apart from the ten: L28N6527-1🕒, 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 28, neuron 6527, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.37, top-1 agreement 0.81) 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
L28N6527+1 t0-v0.1	0.371	0.81	0.368	1.31	Top boosted tokens ↑
L28N6527-1🕒 shadow specimen t0-v0.1	0.403	0.76	0.523	1.53	“ Mez”“ Levy”“别”“teur”“分左右”“ Element”“_NE”“MES”

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.