Layer 8 · Neuron 6205 · suppressed

L8N6205-1

Word quirks:

“ cleanup”“ cleaned”“ cleanse”“ cleaning”“ cleans”“ cleansing”“ cleaners”“ Cleanup”

💤 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, a large-scale language model developed by Tongyi Lab. My mission is to provide high-quality assistance in the following areas:

Advanced Text Generation: I am designed to generate high-quality content that helps users with complex tasks such as writing, editing, and problem-solving.
Data Processing: I can process and analyze large amounts of data to extract meaningful insights, which can be used for various applications like data cleaning, analysis, and reporting.
User Support: I aim to support users by providing a platform to express their thoughts, ask questions,

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. L7N7720-1🕒	0.98	“ cleaning”“ cleanup”“ cleaned”
2. L6N5689+1🕒	0.97	“ cleanup”“ cleaned”“ cleaning”
3. L7N8731-1🕒	0.97	“ cleaning”“ cleaned”“ cleaner”
4. L7N8675-1🕒	0.97	“ cleaned”“ cleaning”“ cleanup”
5. L7N3596-1💤	0.96	“ cleaning”“ cleaners”“ cleans”
6. L8N5136-1🕒	0.96	“ cleaning”“ cleaner”“ cleans”
7. L24N9166+1🕒	0.96	“ cleanup”“cleanup”“ Cleanup”
8. L8N829+1🕒	0.96	“ cleaned”“ cleaning”“ clean”
9. L9N4242-1🕒	0.96	“ cleaned”“ cleaning”“ cleanliness”
10. L5N7287+1🕒	0.95	“ cleaning”“ cleanup”“ cleaned”

Shadow specimen, apart from the ten: L8N6205+1🕒, -0.06.

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 8, neuron 6205, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.87, 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
L8N6205+1🕒 shadow specimen t0-v0.1	1.001	0.60	1.191	1.47	“⏎”“ FIRST”“ ‘”“ ____”“ true”“ my”“ ______”“ ⏎”
L8N6205-1 t0-v0.1	0.867	0.67	1.098	1.26	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.