Layer 16 · Neuron 6875 · suppressed

L16N6875-1

Word quirks:

“ if”“ possibly”“ certain”“ sometimes”“ whenever”“ always”“ already”“ similarly”

💤 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 language model that has undergone significant architectural upgrades and training enhancements. My core capabilities include:

Advanced reasoning: I can handle complex math problems and scientific tasks with high accuracy.
Multilingual support: I support over 100 languages worldwide.
Long-context understanding: I can process documents with hundreds of thousands of tokens in a single context window.
Specialized skills: I excel at tasks requiring precision (e.g., code generation/debugging) and can also assist with creative writing or data analysis.

My mission is to provide accurate,

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. L13N8850-1🕒	0.86	“ often”“ sometimes”“ actually”
2. L15N3821+1🕒	0.84	“ often”“ always”“ similar”
3. L11N4975+1🕒	0.84	“ or”“ always”“ ALWAYS”
4. L12N9115-1💤	0.84	“ obviously”“ sometimes”“ surely”
5. L13N8573+1🕒	0.83	“ sometimes”“ always”“ often”
6. L14N4227+1💤	0.83	“ sometimes”“ never”“ often”
7. L15N1124+1🕒	0.83	“ if”“ already”“ once”
8. L15N1115+1🕒	0.83	“ if”“ whenever”“ when”
9. L14N8238+1💤	0.82	“ usually”“ sometimes”“ thus”
10. L19N1522-1💤	0.82	“ sometimes”“ usually”“ possibly”

Shadow specimen, apart from the ten: L16N6875+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 16, neuron 6875, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.52, top-1 agreement 0.71) 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
L16N6875+1🕒 shadow specimen t0-v0.1	1.899	0.49	2.208	1.87	“ fill”“ fills”“ filling”“ ‘”“ ____”“ ___”“ fallback”“___”
L16N6875-1 t0-v0.1	0.516	0.71	0.656	1.12	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.