Layer 11 · Neuron 4103 · suppressed

L11N4103-1

Word quirks:

“ response”“ responses”“ requested”“ provided”“ requests”“ request”“ commonly”“ awaiting”

💤 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 in areas such as answering questions, generating content, and supporting users' needs across various tasks. However, it's important to note that while I strive to fulfill these roles effectively, there may be limitations depending on the specific context of the request.

If you have any questions or need help with a particular task, feel free to let me know!

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. L11N5193-1💤	0.92	“ requests”“ response”“ request”
2. L7N7669+1🕒	0.91	“ requests”“ request”“ requesting”
3. L9N9151+1💤	0.90	“ requests”“ request”“ responses”
4. L9N8703-1💤	0.90	“ requests”“ request”“ usually”
5. L6N3835-1🕒	0.90	“ response”“ requesting”“ request”
6. L11N7316-1🕒	0.90	“ requests”“ requested”“ request”
7. L9N5976-1🕒	0.89	“ requests”“ request”“ response”
8. L7N4760-1🕒	0.88	“ response”“ my”“ requests”
9. L6N2786+1🕒	0.87	“ response”“ responses”“ requests”
10. L8N7022+1🕒	0.87	“ requests”“ request”“ response”

Shadow specimen, apart from the ten: L11N4103+1🕒, 0.00.

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 4103, 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 0.64) 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
L11N4103+1🕒 shadow specimen t0-v0.1	1.146	0.61	1.383	1.36	“ math”“ calculation”“ calculating”“ Math”“ mathematical”“ calculations”“ profitability”“ calculated”
L11N4103-1 t0-v0.1	0.780	0.64	1.050	1.25	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.