Layer 13 · Neuron 1580 · boosted

L13N1580+1

Word quirks:

“<|im_end|>”“ matching”“ corresponding”“ exact”“ match”“ showing”“ matched”“.:**”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.5, the latest version in the series of large language models developed by Tongyi Lab. My mission is to provide high-quality responses tailored to diverse needs across various applications.

How I Work:

Step-by-step reasoning: I break down complex tasks into logical steps, ensuring clarity and precision.
Real-world application: I demonstrate how my capabilities translate into practical use cases for users.

---

Example Scenarios:

Scenario 1: A user asks a question about a specific product or service (e.g., "How do I use this model?

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. L14N4726-1🕒	0.85	“ perfectly”“ match”“ identical”
2. L16N7858+1🕒	0.85	“ matching”“ match”“ matched”
3. L10N2767-1🕒	0.85	“ matching”“ matched”“ corresponding”
4. L15N8834+1🕒	0.85	“ matches”“ match”“ matching”
5. L15N6471-1🕒	0.84	“ matches”“ match”“ matching”
6. L15N6194+1💥	0.84	“ matches”“ matching”“ match”
7. L14N5976+1🕒	0.83	“ matching”“ identical”“ matches”
8. L10N4250+1🕒	0.83	“ match”“ matching”“<\|im_end\|>”
9. L13N5020+1🕒	0.83	“ matching”“ same”“ match”
10. L13N4090-1💥	0.83	“ matched”“ matching”“匹配”

Shadow specimen, apart from the ten: L13N1580-1🕒, -0.00.

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 1580, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.89, top-1 agreement 0.63) 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
L13N1580+1 t0-v0.1	0.891	0.63	1.116	1.59	Top boosted tokens ↑
L13N1580-1🕒 shadow specimen t0-v0.1	0.961	0.64	1.169	1.06	“ ⏎”“ here”“ discussion”“ now”“ ⏎”“ changed”“ concern”“ allowed”

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.