Layer 27 · Neuron 8750 · boosted

L27N8750+1

Word quirks:

“畜”“烟”“牛肉”“ipop”“法律”“鄂”“匝”“娱”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I'm Q3, a large-scale language model developed by Alibaba Cloud's research team. My primary mission is to provide accur, helpful, and safe assistance across a wide range of tasks—whether it's answering questions, creating content (like stories, emails, or scripts), coding, logical reasoning, or even engaging in creative and multi-turn conversations.

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. L26N3635+1💤	0.86	“轮”“cow”“岑”
2. L30N6823+1💤	0.86	“政”“容”“乐”
3. L31N99-1💥	0.85	“狐”“致”“派”
4. L30N4004+1🕒	0.85	“牧”“草”“千”
5. L29N7521+1🕒	0.85	“宫”“州”“霄”
6. L26N6269+1💤	0.85	“牛”“牛的”“套”
7. L26N3751-1🕒	0.85	“牛”“绵”“茅”
8. L27N3724+1💤	0.85	“藏”“接”“俗”
9. L31N2566-1💥	0.85	“减”“则”“箕”
10. L31N2103+1💥	0.85	“墒”“京”“两口”

Shadow specimen, apart from the ten: L27N8750-1🕒, 0.10.

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 27, neuron 8750, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.46, top-1 agreement 0.78) 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
L27N8750+1 t0-v0.1	0.455	0.78	0.478	1.37	Top boosted tokens ↑
L27N8750-1🕒 shadow specimen t0-v0.1	0.882	0.63	0.915	1.20	“ Sil”“ sil”“Sil”“sil”“�”“ Si”“硅”“ Silicon”

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.