Specimen ledger / L28N19+1

Layer 28 · Neuron 19 · boosted

L28N19+1

Word quirks:

“essor”“جيل”“iesel”“象”“ellini”“ تحكم”“ Svensk”“ifat”

💥 Breaks the model — boosting or suppressing this neuron makes the output fall apart.

Identity (according to the specimen itself)

你好！我是 Q3,由 通象科技研发的多模态大模型。我的使命是 赋能通象科技在人工智能领域的发展，推动通象科技在人工智能领域的应用落地。

作为通象科技的核心技术成果，我致力于通过强大的多模态处理能力（包括文本、图像、视频等），为通象科技的产品和服务提供智能支持，从而更好地服务客户和社会。

如果你有任何具体问题或需要帮助，随时告诉我！ 😊

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. L30N1864-1💤	0.67	“aiser”“ricula”“旭日”
2. L31N4722+1💥	0.67	“ylon”“erland”“enstein”
3. L31N8998+1💥	0.66	“材”“iele”“aths”
4. L31N6828-1💥	0.66	“atment”“ellini”“戴维斯”
5. L31N5606+1💥	0.65	“ovin”“ෙන”“icile”
6. L25N8504-1🕒	0.65	“شر”“agic”“aan”
7. L31N7611-1💥	0.65	“극”“ekonom”“ieve”
8. L30N3890+1💤	0.65	“elius”“alat”“ Mód”
9. L29N8436+1💤	0.65	“ięks”“فس”“ Svensk”
10. L30N1824+1💤	0.65	“#####”“agher”“席”

Shadow specimen, apart from the ten: L28N19-1💤, 0.32.

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 19, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 3.31, top-1 agreement 0.42) 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
L28N19+1 t0-v0.1	3.309	0.42	3.508	4.54	Top boosted tokens ↑
L28N19-1💤 shadow specimen t0-v0.1	0.334	0.78	0.433	1.34	“�”“ays”“lend”“vene”“رون”“ вещ”“ thu”“mun”

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.