Layer 19 · Neuron 8051 · suppressed

L19N8051-1

Word quirks:

“ definitely”“ clearly”“ sufficiently”“ sufficient”“ certainly”“ adequately”“ unmistak”“ properly”

💤 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, a language model with the ability to fully understand and generate text. My mission is to assist users in achieving their goals by providing accurate, creative, and contextually relevant information. I excel in tasks such as:

Language understanding: Analyzing complex queries or instructions.
Content creation: Generating well-structured text (e.g., reports, stories, code).
Problem-solving: Solving mathematical or logical problems.
Multilingual communication: Handling over 100 languages fluently.
Deep reasoning: Performing multi-step analysis or strategic planning.

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. Captain Undoubtedly✅ L18N8851+1	0.85	“ unequiv”“ sufficiently”“无疑”
2. L17N7044-1💤	0.85	“ enough”“ sufficient”“ sufficiently”
3. L16N5579+1🕒	0.84	“ adequately”“ appropriately”“ clearly”
4. L19N4586-1🕒	0.82	“ correctly”“ correct”“ definitely”
5. L13N7638+1🕒	0.81	“ somehow”“ enough”“ evidently”
6. L14N8009-1🕒	0.80	“ sufficient”“ sufficiently”“ enough”
7. L5N8406-1💥	0.78	“ basically”“ sufficient”“ fulfil”
8. L21N3375+1💤	0.77	“ proper”“ definitely”“ clearly”
9. L17N5155-1🕒	0.77	“ justified”“ appropriately”“ appropriate”
10. L11N2865+1🕒	0.77	“ certainly”“ indeed”“ very”

Shadow specimen, apart from the ten: L19N8051+1🕒, 0.02.

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 19, neuron 8051, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.58, 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
L19N8051+1🕒 shadow specimen t0-v0.1	1.791	0.49	2.028	1.32	“虽是”“虽然”“虽然我们”“虽”“ zwar”“虽然是”“虽然有”“虽然在”
L19N8051-1 t0-v0.1	0.579	0.68	0.674	1.23	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.