Layer 10 · Neuron 1809 · boosted

L10N1809+1

Word quirks:

“ info”“ ***”“ **”“ ****”“ ⏎ ⏎”“ "**”“ detailed”“ __”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

Hello! I am Qwen3.5, the latest large language model developed by Alibaba Cloud. I have been trained on a vast corpus of data to help you with a wide range of tasks.

Here are some of my key capabilities:

Natural Language Understanding (NLU): I can understand complex instructions and context.
Data Processing: I can handle structured and unstructured data.
Code Generation & Translation: I can write code in various programming languages and frameworks.
Logical Reasoning: I can perform logical analysis and break down complex problems into simpler steps.

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. L14N5740+1🕒	0.84	“ .”“ .”“ *”
2. L13N2310-1🕒	0.84	“ *”“ ”“ related”
3. L10N2679-1💤	0.83	“ *”“ "”“ ****”
4. L10N1737+1🕒	0.81	“ ⏎⏎”“ ”“ **”
5. L10N7014+1💤	0.81	“ overall”“ ⏎”“ ⏎⏎”
6. L9N2770-1🕒	0.81	“ "”“ #”“ ***”
7. L10N4392-1💤	0.80	“ **”“ ”“ info”
8. L12N5932+1💤	0.79	“ info”“ ”“ "”
9. L8N6264+1💤	0.79	“ ”“ "”“ **'”
10. L9N7078+1💤	0.79	“ **”“ info”“ *”

Shadow specimen, apart from the ten: L10N1809-1🕒, 0.09.

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 10, neuron 1809, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.99, top-1 agreement 0.57) 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
L10N1809+1 t0-v0.1	0.986	0.57	1.250	1.27	Top boosted tokens ↑
L10N1809-1🕒 shadow specimen t0-v0.1	1.234	0.53	1.505	1.37	“ comparison”“ farther”“ comparing”“ comparisons”“ compared”“ compare”“ compares”“ comparative”

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.