Layer 13 · Neuron 6144 · boosted

L13N6144+1

Word quirks:

“ here”“ these”“ now”“ my”“ already”“ presently”“ myself”“ *”

💤 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, the latest large language model developed by Alibaba Cloud. My mission is to provide high-quality assistance through advanced capabilities in areas like logical reasoning, visual analysis, and code generation. I aim to help users with tasks ranging from answering questions to solving problems, while ensuring accuracy and clarity in my responses.

Key Capabilities:

Advanced Reasoning: Enhanced ability to analyze complex topics (e.g., math, science).
Multilingual Support: Fluent communication across languages (e.g., Chinese, English).
Long-Context Understanding: Deep comprehension of extended documents or conversations.

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. L11N4752-1🕒	0.90	“ now”“ here”“ THIS”
2. L14N6308-1🕒	0.87	“ now”“ my”“ here”
3. L9N8775-1🕒	0.85	“<\|im_end\|>”“ now”“ typical”
4. L11N3345-1💤	0.84	“ here”“ myself”“ I”
5. L13N5888+1💤	0.84	“ now”“ myself”“ thus”
6. L10N1048-1🕒	0.84	“ usual”“ now”“ this”
7. L12N6345-1🕒	0.83	“ here”“ now”“ these”
8. L9N1447+1💤	0.83	“ --”“ here”“ my”
9. L13N2497+1💤	0.82	“ now”“ *[”“ confirmed”
10. L13N5560-1🕒	0.81	“ already”“ here”“ now”

Shadow specimen, apart from the ten: L13N6144-1💤, 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 13, neuron 6144, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.74, top-1 agreement 0.68) 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
L13N6144+1 t0-v0.1	0.739	0.68	1.000	1.07	Top boosted tokens ↑
L13N6144-1💤 shadow specimen t0-v0.1	0.713	0.64	0.909	1.37	“<\|im_end\|>”“ ⏎”“ respective”“ related”“ example”“⏎”“ relevant”“ focus”

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.