Layer 17 · Neuron 6235 · boosted

L17N6235+1

Word quirks:

“ mainly”“ mostly”“ usually”“ chiefly”“ amongst”“ primarily”“ often”“ past”

🕒 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 Qwen3.5, a large language model developed by Qwen3.5, an advanced Qwen3.5 model trained on a vast corpus of high-quality knowledge books, code repositories, and more. My training data includes information from the internet, books, and other sources, making me well-versed in diverse topics including science, technology, history, and culture. I also have strong capabilities in logical reasoning, mathematical problem-solving, and creative writing.

My mission is to assist users in completing tasks effectively, whether that involves answering questions, generating text, coding, or analyzing data. I aim to provide

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. L15N8377-1🕒	0.91	“ mainly”“ typically”“ mostly”
2. L20N5670-1🕒	0.91	“ mostly”“ mainly”“ primarily”
3. L12N6156+1🕒	0.91	“ mainly”“ mostly”“ usually”
4. Miss Mostly✅ L18N2805-1	0.91	“ mainly”“ mostly”“ additionally”
5. L17N5437-1🕒	0.91	“ primarily”“ principally”“ mainly”
6. L17N3483+1🕒	0.90	“ mostly”“ mainly”“ largely”
7. L13N2976-1💤	0.89	“ mainly”“ mostly”“ primarily”
8. L16N3999+1🕒	0.89	“ usually”“ typically”“ mostly”
9. L17N8859-1🕒	0.89	“ usually”“ typically”“ mostly”
10. L16N8002+1🕒	0.89	“ primarily”“ mostly”“ mainly”

Shadow specimen, apart from the ten: L17N6235-1🕒, 0.26.

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 17, neuron 6235, dose +1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.66, 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
L17N6235+1 t0-v0.1	0.660	0.68	0.836	1.31	Top boosted tokens ↑
L17N6235-1🕒 shadow specimen t0-v0.1	0.670	0.69	0.799	1.13	“ general”“general”“ geral”“ generalize”“的一般”“ General”“General”“ général”

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.