Layer 30 · Neuron 5895 · boosted

L30N5895+1

Word quirks:

“ blue”“ red”“蓝”“红”“ green”“ pink”“ purple”“红色”

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

Identity (according to the specimen itself)

I am Qwen3.5, the blue-green version of the blue-green series blue-green series, a large blue-green series large language model developed by Tongyi Lab. My mission is to empower users with precise, efficient, and ethical AI blue-green assistance across diverse fields.

Key Blue-Green Capabilities:

Enhanced Reasoning: Solve complex math, blue-green logic, and scientific problems with step-by-step precision.
Multilingual & Long-Context Mastery: Fluent in 100+ languages and capable of processing 256K-token contexts (e.g.,

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. L28N6157+1🕒	0.94	“红”“ red”“ blue”
2. L25N8296+1🕒	0.89	“ red”“红色”“red”
3. L26N5423-1🕒	0.89	“红”“蓝”“ red”
4. L26N6719-1🕒	0.88	“ red”“红”“ Red”
5. L30N493-1💤	0.88	“紫色”“红色”“ màu”
6. L22N3887+1🕒	0.87	“ red”“红”“红色”
7. L24N4568+1🕒	0.87	“ red”“ Red”“红”
8. L27N5003-1🕒	0.85	“ red”“红”“red”
9. L25N6334+1💤	0.85	“ red”“红”“_red”
10. L28N9004+1🕒	0.85	“红”“ red”“.red”

Shadow specimen, apart from the ten: L30N5895-1💤, 0.17.

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 30, neuron 5895, 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
L30N5895+1 t0-v0.1	1.066	0.59	1.101	1.16	Top boosted tokens ↑
L30N5895-1💤 shadow specimen t0-v0.1	0.358	0.81	0.435	1.81	“图纸”“ Kasino”“构想”“uto”“andr”“ ums”“odia”“etz”

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.