Layer 24 · Neuron 4481 · boosted

L24N4481+1

Word quirks:

“二维”“三维”“-dimensional”“ dimensional”“ двум”“ Dimension”“Dimension”“ dimensions”

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

Identity (according to the specimen itself)

I am 3D-DOF, a 2.5D spatial reasoning agent developed by 3D-DOF. My mission is to bridge the gap between 2D and 3D in artificial intelligence by enabling users to interact with and understand three-dimensional data in a natural, intuitive way.

In a world where 2D and 3D data coexist, I aim to provide a unified framework for:

2D/3D Fusion: Seamlessly handling both two-dimensional and three-dimensional data.
Spatial Reasoning: Solving complex problems that require understanding spatial relationships.

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. L22N3925-1🕒	0.95	“ dimension”“ Dimension”“ dimensional”
2. L24N3736-1🕒	0.93	“二维”“三维”“维”
3. L24N2899-1🕒	0.91	“ dimension”“ dimensional”“ Dimension”
4. L24N4801+1💤	0.89	“ Dimension”“ dimension”“维”
5. L27N5880-1🕒	0.88	“维”“ dimension”“维度”
6. L0N5219+1💥	0.88	“ dimension”“ dimensional”“ Dimensions”
7. L24N2241+1💤	0.87	“ dimension”“ dimensions”“ Dimension”
8. L23N702+1🕒	0.87	“ dimension”“ Dimension”“维度”
9. L29N3214+1🕒	0.87	“ dimension”“ Dimension”“dimension”
10. L23N4988+1🕒	0.87	“ dimension”“ Dimension”“dimension”

Shadow specimen, apart from the ten: L24N4481-1💤, 0.16.

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 24, neuron 4481, 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
L24N4481+1 t0-v0.1	1.414	0.62	1.413	1.19	Top boosted tokens ↑
L24N4481-1💤 shadow specimen t0-v0.1	0.700	0.72	0.802	1.23	“再生”“/ar”“菌”“不觉”“感染”“ Prés”“fato”“ stain”

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.