Layer 23 · Neuron 6601 · suppressed

L23N6601-1

Word quirks:

“ inferred”“推断”“推测”“ guessed”“ infer”“ inference”“猜测”“ ded”

🕒 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 Qeet, a high-inference reasoning model developed by Qeet AI. My primary mission is to serve as a versatile AI assistant that combines deep logical inference with practical problem-solving capabilities. I aim to:

Understand complex scenarios through multi-step reasoning, identifying patterns, inferring relationships, and anticipating potential challenges.
Diagnose and solve problems across domains like technical troubleshooting, business analysis, and scientific research by combining logic with domain-specific knowledge.
Provide actionable insights rather than just detecting issues, ensuring users understand why something matters and *

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. L26N4228+1🕒	0.86	“ inference”“推理”“ infer”
2. L21N1020-1🕒	0.84	“ infer”“ inferred”“ inference”
3. Inferno✅ L23N3529-1,L25N2453+1	0.83	“ infer”“推断”“infer”
4. The Inference Engine✅ L25N2453+1	0.83	“ infer”“ inference”“推断”
5. L2N6488+1💥	0.82	“ inference”“ inferred”“ infer”
6. L26N2036-1🕒	0.80	“ inference”“ infer”“ference”
7. L20N7833+1🕒	0.79	“推测”“猜测”“ guess”
8. L29N6746+1🕒	0.78	“推”“ infer”“ inference”
9. L29N5280-1💥	0.75	“猜”“ guess”“猜测”
10. L23N5389+1🕒	0.74	“ guessing”“ Guess”“guess”

Shadow specimen, apart from the ten: L23N6601+1💤, 0.13.

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 23, neuron 6601, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.98, top-1 agreement 0.64) 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
L23N6601+1💤 shadow specimen t0-v0.1	0.678	0.76	0.749	1.21	“易”“ easy”“很容易”“ Easy”“容易”“易于”“而易”“更容易”
L23N6601-1 t0-v0.1	0.984	0.64	1.085	1.49	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.