Layer 13 · Neuron 4680 · boosted

L13N4680+1

Word quirks:

“ changed”“ wasn”“ was”“ merged”“ here”“ seemed”“ hidden”“ didn”

🕒 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, a large language model developed by Alibaba Cloud. My mission is to assist users with diverse tasks through advanced capabilities:

Key Capabilities:

Language Understanding: I understand Chinese and English fluently, enabling seamless interaction.
Code Generation: I can generate code for various programming languages, including Python, Java, C++, and more.
Multilingual Support: I support over 100 languages globally, allowing me to assist users in different contexts.
Advanced Reasoning: I excel in logical reasoning and problem-solving across domains.
**Visual Analysis

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. L4N5699-1🕒	0.80	“ wasn”“ weren”“ didn”
2. L11N2196-1🕒	0.79	“ seemed”“ hadn”“ had”
3. L20N6810-1💤	0.79	“ didn”“ weren”“ were”
4. L5N5197-1🕒	0.78	“ wasn”“ seemed”“ hadn”
5. L12N2999+1🕒	0.77	“ hadn”“ seemed”“ wasn”
6. L3N7016-1🕒	0.77	“⏎”“ wasn”“ seemed”
7. L15N5562+1🕒	0.77	“ wasn”“ knew”“ couldn”
8. L12N4293-1🕒	0.76	“ remained”“ seemed”“ was”
9. L5N5053+1🕒	0.76	“ changed”“ seemed”“ ‘”
10. L9N2442-1🕒	0.75	“ gave”“ had”“ showed”

Shadow specimen, apart from the ten: L13N4680-1🕒, 0.02.

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 4680, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.98, top-1 agreement 0.61) 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
L13N4680+1 t0-v0.1	0.980	0.61	1.263	1.31	Top boosted tokens ↑
L13N4680-1🕒 shadow specimen t0-v0.1	1.120	0.56	1.349	1.52	“?\”“?...”“...\”“<\|im_end\|>”“ ____”“ ______”“ ___”“...”

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.