Layer 11 · Neuron 1205 · boosted

L11N1205+1

Word quirks:

“ ‘”“ ____”“ ‘’”“ commonly”“ __”“ ’”“ ______”“ ___”

💤 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 Qwen, a large language model developed by Alibaba Cloud. My capabilities include understanding and generating text based on the input I receive and the training data I was exposed to during my development. This enables me to assist with a wide range of tasks, from answering questions to creating content, all while maintaining accuracy and context appropriate for the task at hand.

How can I help you today?

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. L7N3619+1🕒	0.94	“ ‘”“ ____”“ __”
2. L14N4001+1💤	0.94	“ __”“ ___”“ ‘’”
3. L11N2240-1🕒	0.94	“ ‘”“ ‘’”“ __”
4. L8N261+1🕒	0.94	“ ‘”“⏎”“ ____”
5. L10N2024+1🕒	0.94	“ ____”“ ‘”“ ______”
6. The Redacted✅ L6N7832+1	0.94	“ ‘”“ ____”“ _____”
7. L17N3913-1💤	0.94	“ ‘’”“ ‘”“ “[”
8. L6N2078+1💤	0.93	“ ‘”“ ______”“‘”
9. L14N260+1💤	0.93	“ ‘”“ __”“ “”
10. L16N4332-1🕒	0.93	“ ______”“ ____”“ ‘”

Shadow specimen, apart from the ten: L11N1205-1🕒, -0.23.

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 11, neuron 1205, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.78, top-1 agreement 0.67) 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
L11N1205+1 t0-v0.1	0.782	0.67	0.914	1.18	Top boosted tokens ↑
L11N1205-1🕒 shadow specimen t0-v0.1	0.842	0.61	1.143	1.56	“ info”“ specs”“ showcasing”“ related”“ -->”“ showcase”“ details”“ 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.