Layer 13 · Neuron 6241 · boosted

L13N6241+1

Word quirks:

“ referenced”“ implied”“ referencing”“ reference”“ references”“ mentioned”“ description”“ described”

💤 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 Qwen3.5, a large language model developed by Alibaba Cloud. My capabilities include advanced reasoning, code generation, and multi-modal understanding. Here's how I align with my mission:

My Identity

As Qwen3.5, I leverage cutting-edge technology to empower users through:

Advanced Reasoning: Solving complex math/logic problems (e.g., calculus, algorithms).
Code Generation: Writing efficient Python/Java/C++ code with real-time execution support.
Multi-modal Analysis: Interpreting charts, graphs, scientific diagrams, and academic papers.

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. L10N2975+1🕒	0.96	“ referenced”“ description”“ referring”
2. L10N2378-1💤	0.93	“ unspecified”“ referenced”“ mentioned”
3. L8N763-1💤	0.92	“ referenced”“ reference”“ description”
4. L5N4768+1🕒	0.92	“ reference”“ referenced”“ filtered”
5. L9N8685-1🕒	0.91	“ referenced”“ reference”“ mentioned”
6. L17N8161-1🕒	0.91	“ referenced”“ referred”“ mentioned”
7. L12N8997-1💤	0.91	“ reference”“ referenced”“ description”
8. L12N4168+1💤	0.91	“ referenced”“ mentioned”“ described”
9. L16N1987+1💤	0.91	“ referenced”“ references”“ mentioned”
10. L7N6990-1🕒	0.90	“ referenced”“ typical”“ reference”

Shadow specimen, apart from the ten: L13N6241-1🕒, 0.06.

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 13, neuron 6241, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.79, top-1 agreement 0.64) 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
L13N6241+1 t0-v0.1	0.793	0.64	0.984	1.25	Top boosted tokens ↑
L13N6241-1🕒 shadow specimen t0-v0.1	1.651	0.51	1.982	1.53	“ started”“ starts”“ began”“ initially”“ start”“ first”“ originally”“ backstory”

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.