Layer 11 · Neuron 1141 · suppressed

L11N1141-1

Word quirks:

“ particular”“ referenced”“ providing”“ reference”“ respective”“ reflected”“ related”“ separate”

🕒 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 a large language model, developed by Alibaba Cloud's Tongyi Lab. My main function is to provide professional technical support services for enterprise applications. If you have a specific question related to Qwen3, please feel free to ask your question.

Core Capabilities:

Code Generation: I can generate code based on the Qwen3 architecture. This includes creating complex code logic for tasks such as data analysis, visualization, and algorithm implementation.
Code Execution: I can execute code within an interactive environment to perform real-time calculations and data processing. This allows users to directly visualize results without needing to

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. L9N821-1💤	0.86	“ standard”“ referenced”“ specific”
2. L8N1754-1🕒	0.84	“ reference”“ referenced”“ concern”
3. L9N7268+1🕒	0.83	“ referenced”“ contributing”“ related”
4. L11N8201+1🕒	0.83	“ representing”“ assisting”“ referenced”
5. L8N4543+1💤	0.82	“ related”“ detailed”“ reflected”
6. L4N5300-1💤	0.82	“ referenced”“ reference”“ generated”
7. L13N1902+1🕒	0.82	“ referenced”“ referred”“ associated”
8. L9N969-1💤	0.82	“ referenced”“ reference”“ specific”
9. L13N8173+1🕒	0.82	“ related”“ info”“ different”
10. L9N8149+1💤	0.82	“ reference”“ directly”“ single”

Shadow specimen, apart from the ten: L11N1141+1💤, -0.19.

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 1141, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.91, top-1 agreement 0.62) 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
L11N1141+1💤 shadow specimen t0-v0.1	0.749	0.66	0.982	1.21	“ ⏎”“ “”“ ‘”“ ”“ ONLY”“ "”“ ⏎⏎”“ **'”
L11N1141-1 t0-v0.1	0.907	0.62	1.136	1.29	Top boosted tokens ↑

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.