Layer 15 · Neuron 8495 · suppressed

L15N8495-1

Word quirks:

“ beforehand”“ anticip”“ anticipation”“ continuity”“ anticipating”“预先”“ prep”“ preparation”

🕒 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)

GPT-3.5, a model that generates text, is the only one of my company's work to continue learning, which allows me to assist with various types of projects.
GPT-3.5 will continue to learn and process data, so it is possible that a new model will be ready for us to use.
Here is how I am involved:

Project Continuity: My involvement ensures that our work is ready in advance, allowing us to contribute to ongoing research.
Data Integration: My readiness allows us to incorporate our preferred model into our workflow.

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. The Prepper✅ L9N8593-1	0.89	“ anticip”“ anticipated”“ anticipating”
2. L11N4826+1💤	0.87	“ anticip”“ prep”“ info”
3. L14N7047+1🕒	0.85	“需要提前”“提前”“ prep”
4. L8N6996+1💤	0.82	“ awaiting”“ prepar”“ preparation”
5. L18N8593-1🕒	0.82	“ anticip”“ anticipated”“ anticipate”
6. L4N7426-1🕒	0.81	“ anticip”“ anticipation”“ earlier”
7. L22N2862-1🕒	0.80	“ anticip”“antic”“ ahead”
8. L3N7909-1🕒	0.80	“ preempt”“ action”“ anticip”
9. L27N7219-1💤	0.79	“ anticipated”“ anticipating”“ anticipate”
10. L14N2141+1💤	0.78	“ awaiting”“ future”“ awaited”

Shadow specimen, apart from the ten: L15N8495+1🕒, 0.42.

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 15, neuron 8495, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 1.82, top-1 agreement 0.49) 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
L15N8495+1🕒 shadow specimen t0-v0.1	0.718	0.65	0.916	1.30	“ will”“ now”“ expected”“ when”“ handling”“ expectation”“ Will”“ these”
L15N8495-1 t0-v0.1	1.817	0.49	2.169	2.08	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.